Nuclear Facilities Targeted

Russia’s full-scale invasion of Ukraine in February 2022 has centred concerns on the threat that militaries pose to nuclear installations. While Ukraine’s situation is unique, the threat of military strikes on nuclear facilities predates the war in Ukraine, with military force having been used – or seriously considered – against and in the vicinity of nuclear installations on multiple occasions before that. As more countries become interested in pursuing nuclear energy as a source of clean and sustainable power, and as non-proliferation norms are increasingly challenged, the probability that nuclear installations will find themselves the targets – or unintentional victims – of the use of military force could increase. Such military activity should not be normalised; however, political and military leadership must be ready to anticipate, mitigate and respond to potential future military attacks on nuclear installations.

This paper aims to understand the operational and strategic logic of why states may use military force against nuclear installations, as well as the consequences of such attacks, with the intention of identifying approaches for reducing related risks. It presents five contexts in which military force may be used or threatened to be used against nuclear installations. These are: as a counterproliferation tool; for energy disruption purposes; for area denial; to generate escalatory, deterrent or coercive effects; and when a nuclear site is encountered on an axis of advance or during a broader military operation.

Military force has been used or seriously considered against nuclear installations on numerous occasions in the pursuit of counterproliferation aims. This is despite historical cases pointing to significant limitations of military attacks as an effective counterproliferation tool. The use of military force in a counterproliferation context may seek to achieve a range of objectives – from the elimination of an adversary’s programme altogether, to limiting its progress, or simply as a way of sending a signal of displeasure and resolve to counter the perceived threat. Depending on the proliferation pathway a state may be pursuing, counterproliferation strikes may be carried out against a range of installations across the nuclear fuel cycle, each presenting unique risks to civilian populations and the environment.

Russia’s military advance on and occupation of Ukraine’s Zaporizhzhia nuclear power plant, in contrast, appear to have been driven, at least in part, by an intention to disrupt and eventually co-opt the energy generating capacity of the plant. The targeting of enemy energy generation and distribution infrastructure is well established in the military doctrine of many alliances and states – including those of NATO, Russia and China. The expected growth of the importance of nuclear power in the global energy mix in the coming decades may increase the likelihood that future armed conflict will see greater targeting of nuclear energy infrastructure. This may not necessarily entail direct attacks on nuclear reactors but rather assaults on auxiliary systems – such as water and electricity supply to nuclear power plants – under the erroneous assumption that such attacks carry limited risk to nuclear safety.

Attacks on nuclear facilities may also be carried out with the express intention of releasing radioactive or other toxic material as an area denial strategy. Radiological or chemical contamination of territory complicates and delays an adversary’s military operations in the affected area. However, most militaries have at least some ability to shield against CBRN (chemical, biological, radiological and nuclear) hazards and to continue operations in a contaminated environment – particularly if a mission is critical. As such, ultimately, the effectiveness of radiological or chemical contamination for area denial is likely to be limited. Furthermore, the unpredictability of the dispersion of a contaminant from an attacked nuclear facility creates significant risks of contamination to friendly populations and forces and this contamination might expand and escalate a conflict if toxic material reaches the territory of a third country.

Due to the psychological salience of radiological hazards among the general population, attacks and threats of attack on nuclear installations may be used for escalatory, deterrent or coercive purposes. States may threaten attacks on nuclear facilities and the release of radioactive material from these installations as a “half-step” between the use of conventional weapons and a nuclear weapons attack, despite the differences in effect and normative significance of nuclear safety incidents and nuclear weapons attacks.

Short of attacking nuclear installations as a primary target of a military operation, militaries may also encounter and need to contest nuclear facilities on an axis of advance during a land invasion or may be required to carry out air campaigns against territory where nuclear sites are present. Militaries – supported by relevant national authorities – must therefore ensure they have the necessary expertise and capabilities and have conducted the necessary operational planning to be able to safely operate around nuclear facilities if this becomes necessary.

Based on analysis of operational and strategic drivers for the use of military force against nuclear installations, the paper proposes a number of considerations and recommendations to manage and mitigate related nuclear safety risks.

These include:

• Strengthening understanding of, and compliance with, existing international legal principles and norms that restrict the use of military force against and near nuclear installations. This may include developing additional guidance on the operationalisation of existing legal provisions and considering how legitimate military objectives may be pursued while maintaining maximum respect for nuclear safety. Reiterating commitments to existing legal provisions and broader norms, and calling out states that do not respect these, is important for making sure that the use of military force that threatens the safety of nuclear installations and poses a risk to civilians is not normalised.

• Considering options for strengthening passive defences around nuclear installations and auxiliary systems and equipment which support the safe operation of nuclear facilities (including water supply systems and electrical infrastructure). Passive defences may need to be supported with active defence measures in the instance of an active conflict, including the deployment of layered and distributed air defence, as well as specialised C-CBRN (counter-CBRN) ground capabilities. The introduction of any passive and active defences needs to be balanced against limited resources and other priorities that are likely to emerge during an active conflict and should be commensurate with assessed risk in a given context. The deployment of military capabilities near to nuclear installations may also have the unintended consequence of increasing the risks to facilities.

• Creating redundancy in and decentralising nuclear installations and related network. This may help shift the cost–benefit calculus for an attacking state by increasing the number of targets that need to be attacked to have meaningful effects on a system and reducing the effects of single strikes. In the context of nuclear energy generation, this may include a shift away from large gigawatt-scale reactors towards the deployment of small modular reactors. However, decentralisation may have the unintended consequences of increasing the risks of hazardous material release and harm to civilian populations by increasing the number and distribution of facilities containing toxic materials.

• Improving whole-of-society resilience to nuclear safety incidents, including through efforts to raise levels of emergency preparedness, risk awareness and education on nuclear safety among the public and political leadership. This should involve the pre-emptive formulation of risk and crisis communication strategies, as well as tools for countering disinformation and misinformation, which prioritise the establishment of trust between the population and authorities, and offer clear information and instructions. Any such efforts must be commensurate to the risk facing a given state and population, and will have to be balanced against the need to avoid alarmism and the risk of inciting unnecessary anxiety.

Introduction

States have used force against nuclear programmes for nearly as long as humans have known how to split the atom. Between 1942 and 1943, Allied forces carried out a series of operations against the Vemork heavy water production plant in Norway. They suspected – rightly – that the facility was part of a German effort to build a weapon “a thousand times more potent than any in use to-day”. The destruction of the plant and the stocks of heavy water that it was producing thus took the highest priority. The facility was ultimately destroyed in a bombing raid in November 1943.

Since those attacks, states have directed military force against other states’ nuclear programmes on multiple occasions. Iraqi strikes on Iran’s Bushehr Nuclear Power Plant (at the time still under construction) during the Iran–Iraq War, the Israeli attacks on Iraq’s Osirak reactor in 1981 and Israel’s bombing of Syria’s Al-Kibar reactor in 2007 are some of the better-known cases. In numerous other instances, attacks were considered but not carried out, including: Egyptian and Soviet deliberations over striking Israel’s nuclear facility at Dimona (during the 1960s); US as well as Indian and Israeli plans, in the 1970s and 1984 respectively, to attack Pakistan’s nuclear facilities at Kahuta; and US plans under multiple administrations for military strikes against North Korea’s nuclear programme. Russia’s plans for a potential conflict with Japan and South Korea also included plans to target the countries’ nuclear energy infrastructure. Meanwhile, Iran’s continued advancement of its nuclear programme and concerns from the US, Israel and others over its weaponisation potential have once again brought to the fore the debate over the utility of military force for addressing the threat of nuclear weapons proliferation.

Counterproliferation objectives have historically been a prevalent driver for the use of military force (and its serious consideration) against nuclear facilities. However, it has not been the only one. Russia’s military seizure and subsequent occupation of Ukraine’s Zaporizhzhia nuclear power plant (ZNPP) in 2022 have highlighted the broader range of risks posed to nuclear infrastructure from military activity. As nuclear power is expected to become an increasingly important source of energy globally, as conventional conflicts persist in multiple parts of the world, and as non-proliferation norms continue to be challenged, it is likely that military threats to nuclear facilities will persist.

This paper is a contribution to ongoing work on identifying ways to reduce the risks that military force can pose to nuclear installations. It aims to supplement existing analysis – which has, in large part, focused on ways to strengthen legal prohibitions against attacks on nuclear installation and to improve nuclear safety and security measures in times of crisis – by distilling the strategic and operational logic of the use of military force against nuclear facilities. In other words, the paper poses the questions: What may cause militaries to target nuclear facilities? And what does that mean in terms of reducing the risk to nuclear installations posed by military operations? While such military activity should not be normalised, it is important that those working to reduce risks to nuclear sites understand its logic and the risks it may present.

Methodology, Sources and Limitations

This paper builds on existing literature on military threats to nuclear facilities and nuclear safety in conflict zones. It draws on, among others, Bennett Ramberg’s seminal monograph Nuclear Power Plants as Weapons for the Enemy: An Unrecognised Military Peril (first published in 1984), the work of Matthew Fuhrmann and Sarah E Kreps on military attacks on nuclear facilities as a counterproliferation tool, and research and analysis on nuclear safety in conflict zones and related international legal frameworks conducted by the Stockholm International Peace Research Institute, the Nuclear Threat Initiative, the Vienna Center for Disarmament and Non-Proliferation, and RUSI. This paper aims to update and complement some of the analysis presented in these and other works. In part, it does so by integrating lessons learned from Russia’s occupation of Ukrainian nuclear sites and its broader assault on Ukraine’s energy system.

Part of the research for this paper was conducted during field study in Ukraine in July and August 2024. This aimed to capture and learn from Ukraine’s experience in managing the safety and resilience of its nuclear installations and the broader energy grid since the start of Russia’s full-scale invasion. The research included interviews with Ukrainian energy and nuclear safety experts, nuclear operators, nuclear plant workers, legislators and military personnel in Kyiv, Zaporizhzhia and the Chornobyl Exclusion Zone (CEZ). In particular, the fieldwork helped to inform some of the analysis presented in this paper related to the targeting of nuclear facilities for energy disruption, for coercive purposes and when encountered on an axis of military advance. The research also relied on a review of other historical cases of military attacks on nuclear facilities to inform the analysis. Cases were examined in varying degrees of detail, due to time limitations and discrepancies in the availability of information. Cases were identified and studied through review of existing literature, archival documents, and interviews with subject matter experts (SMEs). A list of the cases considered for this research is included in the Annex; this is unlikely to be a comprehensive list of all instances in which nuclear installations were targeted or found themselves in the vicinity of military operations. Furthermore, the precise reason for the use or threat of use of military force against nuclear infrastructure in individual cases was not always possible to ascertain, due to sometimes limited information on an attacking state’s decision-making and the presence of confounding drivers and factors. However, the review of cases allowed for a general sense of the likely motivations, desired effects and other considerations that appear to have driven past attacks. The identified drivers and other relevant considerations were subsequently tested, refined and expanded through semi-structured SME interviews and the review of primary and secondary sources. This included consultation of materials on: national risk and resilience assessments; military targeting doctrine; energy grid design and resilience; chemical, biological, radiological and nuclear (CBRN) hazards and emergency response; international law relating to attacks on nuclear infrastructure; and nuclear safety and security regulation.

Box 1. Legal Considerations and Limitations

While the paper generally foregoes a legal analysis of military attacks on nuclear facilities in various circumstances, it is worth stressing that such military activity exists in highly contested and nuanced legal territory.

The use of military force against civilian infrastructure is generally prohibited under international humanitarian law (IHL), although exceptions may be made if the target is determined to be contributing significantly to an adversary’s military operations and if an attack is deemed militarily necessary. However, the bar for determining military necessity is set high and attacks still need to abide by principles of IHL that require – among other considerations – the effects on civilians of an attack to not be disproportionate to the military advantage being sought. Attacks that cause undue harm to civilians or are intended primarily to terrorise the civilian population are prohibited under international law.

The dual-use nature of many of the activities and materials in a nuclear fuel cycle can complicate the distinction between civilian and non-civilian sites. In a broader military conflict, discerning the extent to which nuclear power installations support military activity can also be difficult. The presence of toxic materials at many nuclear installations makes it difficult to argue that attacks on those sites will not pose a serious risk to civilians and the environment. Yet, the likelihood and extent of the risk and its proportionality in relation to the expected military effects and resulting military advantage can be difficult to assess.

The reliance on individual states’ interpretations and applications of IHL also means that – in instances where a state has decided that another state’s nuclear installations pose a critical threat – the state will likely find ways of justifying the use of military force against a nuclear site. The desire to preserve the use of military force against nuclear installations as a policy option – namely, in the pursuit of non-proliferation objectives – has prevented the introduction of more extensive legal prohibitions and is likely to do so in the future. Russian behaviour in Ukraine also demonstrates that some countries will remain indifferent to their IHL obligations.

As such, while finding ways to strengthen compliance with existing legal restrictions should be part of the effort to minimise risks, measures to make these facilities less attractive targets, to reduce the consequences of any future attacks and to prepare militaries to operate safely around these facilities are equally important.

The author conducted semi-structured research interviews with more than 30 SMEs and practitioners (including military personnel and government officials) by telephone, video conference and in-person between July and November 2024. Some responses to questions were also collected by email. The interviewees’ areas of expertise spanned the fields of nuclear safety and security, nuclear power plant operations, nuclear non-proliferation, strategic security dynamics, nuclear risk reduction, energy grids and energy security, military targeting practices and other military doctrine, military engineering, CBRN emergency preparedness and response, and radiological medicine. Interview questions were tailored to the individuals’ area of expertise and experience and therefore varied considerably across the sample. Interviewees had the option of remaining anonymous or being identified. Those who preferred to remain anonymous have been assigned an alpha identifier alongside a general description of their role and/or area of expertise.

Interviewees were selected based on their areas of expertise and experience and were identified through the author’s professional networks. The sensitive nature of the subject matter and its direct relevance to countries’ national security imposed limitations on the information that could be shared during interviews on matters pertaining to vulnerabilities and defence of nuclear infrastructure. The author was also limited somewhat by geopolitical constraints in the perspectives that could be captured for this research: attempts were not made, for instance, to identify Russian or Chinese interviewees to query their countries’ perspectives on the targeting and defence of nuclear sites, as such attempts were likely to prove fruitless in the current geopolitical climate. Insights into these countries’ attitudes were drawn, where possible, from public statements, policy documents and observable behaviour – but were ultimately limited in this analysis. As such, the sample of experts interviewed for this paper is not representative of the full range of perspectives on the subjects discussed.

Scope and Definition of Concepts

The terms “nuclear facilities”, “nuclear sites”, “nuclear installations” and “nuclear infrastructure” are used interchangeably throughout the paper to refer to “nuclear facilities” in the broadest definition offered by the International Atomic Energy Agency (IAEA) Nuclear Safety and Security Glossary: “A facility (including associated buildings and equipment) in which nuclear material is produced, processed, used, handled, stored or disposed of”. The paper also considers attacks on sites that are meant to host nuclear material but to which nuclear material has not yet been introduced; this would have included Iraq’s Osirak reactor at the time of the 1981 Israeli attack as well as Iran’s Bushehr nuclear power plant, which was under construction when it was repeatedly struck by Iraqi forces. Nuclear facilities vary significantly in their purpose, construction and levels of protection, as well as the radioactivity and other hazards they have the potential to release if they are damaged. As such, where analytically necessary, this paper discusses considerations related to specific types of facilities (for example, nuclear reactors, uranium enrichment plants, spent fuel storage facilities, and so on).

Military attacks also vary widely in their scale, the methods they employ and the context in which they are carried out – including whether they take place as part of a larger military operation (for example, a ground invasion or a protracted air operation) or as one-off strikes outside a broader state of war. The term “military strike” is used in the paper to refer to a hit (or intended hit) on a target using military force – irrespective of the kind of military force used, whether the strike took place in the context of a broader military operation or was a one-off hit, or the ultimate effect on target. The term “military attack” is used to refer broadly to military operations aimed at or conducted near nuclear facilities (which may not include a direct strike on a facility). Defining what constitutes “use of military force” and distinguishing it from other “sub-threshold” activities that states may use against an adversary is challenging and the subject of academic and legal debate. For the purposes of this paper, reviewed cases and analysis were limited to instances when kinetic action was conducted by a state’s military. The paper therefore excludes instances of sabotage operations, cyber attacks, assassinations or intelligence-collection activities. The logic, practicalities, effects and countermeasures for these types of activities are likely to differ considerably from attacks using military force – even if carried out in the context of a broader military operation. As such, these activities deserve separate consideration.

The focus of this paper is the use of military force by states and therefore this paper does not consider attacks against nuclear facilities by non-state actors. The non-state threat has received much greater attention – analytical, regulatory and legal – in the nuclear safety and security space, compared with that given to state threats. Furthermore, the motivations, resources and constraints of non-state actors looking to attack a nuclear facility are likely to differ significantly from those of state militaries. As such, they require separate consideration. As with defining “use of military force”, drawing a clear line between state and non-state actors can sometimes be difficult. For the sake of clarity, this paper has limited itself to an analysis of attacks carried out by state militaries. Nevertheless, state-sponsored non-state actors have threatened nuclear infrastructure in the past and may do so in the future. The interplay between state and non-state motivations, resources and restrictions, and the impact on approaches to defence against such attacks is worth exploring in future research.

I. Drivers and Effects of Military Attacks on Nuclear Infrastructure

In certain contexts, the driver for targeting nuclear infrastructure may be much the same as that for military attacks on other forms of critical national infrastructure (CNI) – for example, the interruption or seizure of energy production (to impede military industry and/or to degrade a population’s morale and will to resist). Nuclear facilities may also be targeted for their intrinsic economic value, just like any other part of large-scale industrial or other infrastructure. However, nuclear facilities have a number of particular characteristics which may present unique strategic and operational opportunities and challenges that influence states’ thinking on whether and how to carry out military activity against or near nuclear infrastructure.

The most relevant particularities of nuclear facilities as a target set are the dual-use nature of some nuclear technology and the presence of material at some nuclear sites which – if dispersed – may pose a radiological danger to human health and the environment. The threat posed by radiological release presents the potential for both physical and psychological harm. Meanwhile, when nuclear sites form part of a nuclear weapons programme – or are suspected of contributing to a state’s nuclear weapons ambitions – they are also likely to be imbued with a level of symbolic significance. Such significance relates to a state’s perception of its own place on the world stage, technical prowess and strategic security, and can exist independently of the actual tactical or operational value of the facilities and the capabilities they may host. The relatively high baseload of nuclear power plants – compared with other energy production facilities – can also present a much larger and centralised energy disruption target for attacking militaries.

The above-listed characteristics of nuclear facilities as targets are the departure point for the analysis that follows of why and how states may carry out attacks against nuclear infrastructure. These characteristics of nuclear facilities, and related consequences of attacks, may be key to why these sites are targeted. For instance, a nuclear installation may be attacked because of, rather than despite, its dual-use nature or the likelihood of radioactive release, as this may be the desired strategic or operational effect. Alternatively, these characteristics and consequences may act as undesirable or unintended consequences of attacks, which may or may not deter or make less attractive attacks on nuclear sites. For example, strikes on other parts of an electrical grid may be carried out despite the risk of creating a nuclear accident at a power plant and the consequent release of radioactive material; alternatively, strikes on the grid may be deterred when the collateral risks are determined to be excessive. This paper attempts to capture both these dynamics in analysing how these unique characteristics have an impact on the operational and strategic considerations for attacks on nuclear sites.

This chapter discusses five overarching operational and strategic drivers of military attacks, and threats of attack, on nuclear facilities: counterproliferation; energy disruption; the release of radioactive material for area denial and defence; escalation of violence in the course of a conflict, or the threat of doing so for deterrent or coercive purposes; and encountering nuclear facilities on an axis of advance or during a broader military operation.

These broad categories were articulated based on a review of historical cases of attacks on nuclear facilities and data collected through research interviews, in which SMEs were asked to consider the range of possible drivers for military attacks on nuclear sites. The drivers identified in this paper are unlikely to be exhaustive and are not mutually exclusive; a decision to attack may be informed by several of these factors simultaneously. The following discussion aims to be a starting point for policymakers, militaries and international organisations looking to understand why militaries may consider attacking nuclear infrastructure and when or how influence can be exerted to reduce risks.

The chapter first provides an overview of the effects of radioactive and other toxic material release from nuclear installations, before moving on to a discussion of some drivers of military attacks on nuclear facilities.

Effects of Nuclear Material Release

The nature of the materials released from nuclear facilities and their practical impact on human health and the environment depend on many factors. These include, but are not limited to: the nature of the material at the facility; the extent and effectiveness of material containment; the quantity of material released; meteorological conditions (such as wind speed, wind direction and precipitation); the availability of protective measures and emergency response; proximity of population centres; and the nature of the environment in the vicinity of the attacked facility (including the presence and type of vegetation and animal life, water sources and the type of soil). Furthermore, nuclear facilities can contain a range of radiotoxic materials, comprising various nuclides and isotopes. Differences in the types of radioactivity the materials release, their half-lives and masses dictate the nature of the threat posed by the various nuclides that may be released from a given facility as a result of an accident or strike.

Box 2. Existing Prohibitions on Military Attacks Against on Nuclear Facilities

The state of current prohibitions on the use of military force against nuclear facilities, as well as the historical trajectory of efforts to strengthen protections against such attacks, have been outlined at length elsewhere. While a detailed summary is not provided in this paper, it is worth briefly outlining the key parameters (an inexhaustive list) of existing international and multilateral protections from military attacks on nuclear installations:

• Geneva Conventions (Additional Protocols I and II): Various provisions of the Geneva Conventions require militaries to take the necessary precautions to avoid harm to civilians; these include, but are not limited to, prohibitions on attacks against facilities containing “dangerous forces” (including nuclear electrical-generating stations).

• IAEA General Conference: While not legally binding, the IAEA has adopted several General Conference resolutions (in 1981, 1983, 1985 and 1990) expressing concerns over military strikes on nuclear facilities. In a 2009 General Conference decision, the IAEA noted that armed attacks against peaceful nuclear facilities constitute “a violation of the principles of the United Nations Charter, international law and the Statute of the Agency”. In September 2023, the IAEA General Conference adopted a resolution expressing concern over Russian military activity at the ZNPP and reiterating the importance of ensuring nuclear safety in armed conflict.

• UN Security Council Resolutions: UN Security Council Resolution 487 (1981) condemned Israel’s attack on Osirak, calling it a violation of the UN Charter. The resolution did not introduce any new binding prohibitions. In July 2024, the UN General Assembly adopted a resolution recalling the 2009 IAEA General Conference decision and condemning the Russian seizure of the ZNPP.

• Nuclear Non-Proliferation Treaty (NPT) Review Conferences: The 2010 NPT Review Conference action plan called for member states to comply with the 2009 IAEA General Conference decision (see above). The 2022 NPT Review Conference draft outcome document (not ultimately adopted) also expressed “grave concern” at attacks and threats of attack on peaceful nuclear facilities.

• IAEA’s Seven Indispensable Pillars and Five Concrete Principles of Nuclear Safety and Security: In response to the threats to Ukraine’s nuclear facilities following Russia’s full-scale invasion, IAEA Director General Rafael Grossi put forward the Seven Indispensable Pillars of Nuclear Safety and Security, drawn from existing IAEA guidance and standards on nuclear safety and security. In May 2023, Grossi also articulated Five Concrete Principles of Nuclear Safety and Security at the ZNPP, tailored specifically to address the risks facing the plant.

There are also regional and bilateral prohibitions on the use of military force against nuclear facilities:

• India–Pakistan Non-Attack Agreement (NAA): The 1988 NAA commits India and Pakistan to refraining “from undertaking, encouraging or participating in, directly or indirectly, any action aimed at causing the destruction of, or damage to, any nuclear installation or facility in the other country”. The Agreement also requires the two countries to exchange information on the location of their nuclear installations on a yearly basis.

• Treaty of Pelindaba: The treaty, which was concluded in 1996 and entered into force in 2009, established a nuclear weapon-free zone on the African continent and prohibits parties to the treaty from carrying out any attacks against nuclear installations in the zone.

Facilities at the front end of the nuclear fuel cycle – namely, uranium mines and mills, uranium conversion sites, uranium enrichment plants and nuclear fuel production facilities – present primarily a chemical rather than radiological risk. Natural uranium and unirradiated uranium compounds (triuranium octaoxide, uranium dioxide and uranium hexafluoride) are not significant sources of radioactivity (with enriched uranium hexafluoride slightly more radioactive than natural uranium). However, these materials can pose a serious chemical risk, primarily from inhalation. For instance, uranium hexafluoride reacts with water vapour present in the air to produce the highly toxic uranium fluoride and, through further reaction, uranyl fluoride, another toxic substance. The threat posed to life from these chemicals again depends on a range of factors, with the most serious effects likely to be fairly localised – although dispersal of some material through wind or groundwater contamination may produce some hazard for populations further away from the site of release.

Release from facilities involved in the service period and back end of the nuclear fuel cycle – namely, nuclear reactors, spent-fuel storage sites and spent-fuel reprocessing facilities – pose significant risk of radiological harm, due to the composition of the irradiated nuclear fuel at these facilities, which contains – in addition to uranium and plutonium – unstable radionuclides especially harmful to human health, which are not present in fresh (unirradiated) fuel. The precise isotopic composition of the irradiated fuel at a given facility (and therefore the nature and volume of harmful isotopes present) depends on the initial make-up of the fuel, its burnup, and, if the fuel is no longer in an operational reactor, how long it has been cooling. As with the chemical hazard discussed above, the extent of the radiological hazard depends on (among other factors) the nature of the facility in question (including the extent of physical protection afforded to the radioactive material), the facility’s operational state at the time of the incident, the nature of the incident itself, the abilities and responsiveness of the facility’s personnel and emergency responders, the proximity of the facility to population centres, and the nature of the ecosystem in the vicinity of the site. The consequences of incidents can vary widely depending on these factors.

The extent to which radioactive material is physically protected from any direct strike or dispersal has an important impact on the likelihood and extent of dispersal following various safety incidents or the use of various weapons systems and munitions. Most nuclear installations benefit from reinforced containment structures or vessels. As discussed in greater detail in the section of this paper on strengthening passive and active defences of nuclear facilities, these protections are meant to protect nuclear installations against a range of threats and to prevent release of hazardous material in case of an accident. However, facilities are not generally designed to withstand military assault or operate under wartime conditions (with the exception of some facilities which may be part of an effort to develop nuclear weapons and/or are expected to face serious military threats from adversaries – for instance, Iran’s underground enrichment sites). The extent of protections also varies across countries and facilities. For instance, many of the Soviet-built reactors still operating in Russia are not under a reinforced concrete containment structure; this includes the reactors at the Kursk nuclear power plant, which caused some concern as Ukrainian armed forces advanced into Russia’s Kursk oblast in summer 2024.

Physical and Psychological Effects

Depending on the amount and nature of the exposure to various kinds of nuclear material, effects on the human body will vary. Acute exposure to significant amounts of radiation may result in acute radiation syndrome (ARS), affecting the bone marrow, the gastrointestinal system, and the cardiovascular and central nervous systems. ARS can generally be expected at a dose of 700 millisieverts (mSv) (the equivalent of receiving about 70 full-body CT scans, or 140,000 dental X-rays, or eating 7 tonnes of Brazil nuts). Symptoms vary in severity and duration and may include – depending on the severity of exposure – nausea and vomiting, fever, confusion, haemorrhaging and even death. Cutaneous radiation injury may also result in damage to skin and hair loss. Changes in blood cells can be readily observed at lower doses, of about 100 mSV (the equivalent of 10 full-body CT scans, 20,000 dental X-rays or 1 tonne of Brazil nuts).

In practice, the impact of radioactive release on human health and the environment – especially long-term effects – can be notoriously challenging to track; the effects of any excess radiation may develop over years and decades and may be challenging to isolate from confounding acute and long-term factors. The data on the impact to human health of the 1986 Chornobyl and 2011 Fukushima-Daiichi accidents (which tend to serve as worst-case scenario references in public commentary) is largely inconclusive – with some exceptions. According to the UN Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) – whose findings have been contested by other sources – of the 600 workers on the site at the time of the Chornobyl accident, 134 suffered from radiation sickness, 28 of whom died in the first three months after the accident. Some subsequent deaths may also be attributable to radiation effects. A marked increase in thyroid cancer among children in Belarus, Ukraine and affected parts of Russia, as well as increased rates of leukaemia and cataracts among those workers that had suffered from radiation sickness, has been recorded in the decades since. However, the UNSCEAR report notes that significant changes in other rates of cancer have not been observed, and a general upward trend in mortality across the area of the Soviet Union and improvements in reporting make it difficult to isolate and accurately assess the long-term health impacts of the accident. After the Fukushima-Daiichi accident, no acute health effects attributable to radiation exposure were immediately reported. The UNSCEAR report on the impact of ionising radiation following the accident did not expect a discernible increase in most cancer occurrences among the Japanese population, although it did note the possibility of increased thyroid cancer rates and a potential (likely not discernible) increase in occurrences of leukaemia among children.

Environmental impacts of the accidents have also been debated. The CEZ – which stretches out to a 30 km2 radius around the Chornobyl nuclear power plant – remains contaminated and largely without permanent inhabitants. Yet, the site remains operational and – prior to the full-scale Russian invasion of Ukraine – was open to public visitors, with remaining radioactive isotopes “at tolerable exposure levels for limited periods of time”. Mutations and other adverse effects on animal and plant species in the CEZ were reported in the years after the accident. Today, the area has been rewilded and become a wildlife sanctuary; however, there is also data to suggest that species diversity remains impacted by radiation. Despite the spread of radioisotopes from the accident to other parts of Europe and subsequent contamination of animal products and plants, the increased exposure of radiation is not believed to have had a direct impact on cancer rates outside the affected Soviet republics. At Fukushima-Daiichi, some areas around the plant remain exclusion zones; many have been deemed safe enough for residents to return.

Assessing the consequences of a given radiological release is particularly challenging for long-term and generational effects of radiation exposure, but this can be true for immediate effects as well. In an interview with the author, Robert Peter Gale – an expert on radiation health effects, who was a leading figure in the medical responses to the 1986 accident at Chornobyl, as well as the 1987 radioactive incident in Goiânia, Brazil, and the Tokaimura and Fukushima-Daiichi accidents in Japan – noted that, immediately following an accident, for victims in close proximity to the event, it may be difficult to tell apart the effects of radionuclide exposure from those of regular burns or exposure to chemicals also released in the accident. This complicates identifying the most appropriate course of immediate treatment for victims.

Besides the physical impacts on human health, concerns over exposure to radioactive materials – whether harmful exposure actually took place or not – can have significant psychological effects on populations. The particular propensity of CBRN incidents to incite fear among the public is a well-studied phenomenon and was raised repeatedly by experts interviewed for this project. Interviewees contrasted the fear of an attack on a nuclear site with reactions to attacks on other energy and strategic targets, such as coal and hydro plants, which do not tend to generate the same levels of panic among the general population. Scholars who have studied the subject of risk communication in response to CBRN incidents (namely, CBRN terrorist attacks) have noted that “CBRN releases score highly on all characteristics that trigger fear and anxiety”. The factors that make CBRN a particularly anxiety-inducing threat include relatively low levels of understanding of nuclear technology and the effects of radioactivity, the ability of CBRN incidents to induce “severe and unusual medical conditions”, the unpredictability of the threat, and its manufactured nature (as opposed to “acts of God”). As some of the experts interviewed for this paper pointed out – and as has been well-documented in literature – the public anxiety caused by radiological release is likely to be incommensurate with the actual physical danger. This may, in turn, create additional challenges and risk as the “worried well” overwhelm emergency response services and potentially engage in unnecessary and possibly harmful remedial action (for instance, unnecessarily evacuating their homes or self-medicating).

In his conversation with the author, Gale pointed to the panic that spread to Kyiv and Tokyo after the Chornobyl and Fukushima-Daiichi disasters respectively, despite the fact that the disaster sites were a considerable distance away from these cities and, in Gale’s assessment, did not pose a significant radiation threat to the cities’ populations. Instead, he noted that the subsequent evacuations from surrounding areas, as well as precautionary measures taken out of fear of the effects of radiation exposure, may have caused greater collateral harm. Gale pointed to the increase in induced abortions after the Chornobyl accident, as a result of misguided advice from doctors on the effects of the radiation on foetuses. Studies conducted in the years following the disaster found no evidence that children exposed to radiation in utero showed any neuropsychological differences from the control sample.

Evacuations following a nuclear accident can also have harmful secondary effects on the displaced population. Gale and others have highlighted the risks of disruptions to healthcare, particularly for the older population, caused by evacuations from around the Fukushima-Daiichi plant after the 2011 disaster. Such evacuations appear to have had a greater direct health impact than radiation exposure from the accident. In its report on the levels and effects of radiation exposure after the accident, UNSCEAR reported on the “immediate aggravation of the condition of already vulnerable groups: for example, more than 50 hospitalized patients were reported to have died either during or soon after the evacuation … upwards of 100 elderly people may have died in subsequent months because of a variety of conditions linked to the evacuation”. The influx of people into neighbouring regions and countries also has important social, political and health implications as public services, including healthcare, must stretch to accommodate the new arrivals.

The intermediate- and long-term psychological effects of radiological release incidents have also been well-documented. Studies have found increased levels of depression, anxiety and “medically unexplained physical symptoms” among survivors of the Chornobyl disaster than in control samples. A 2006 World Health Organization study also reported that distress among victims was also often the result of subjective perceptions that they, or their children, had been affected by the radiation release at Chornobyl. The study noted expert consensus that, “The mental health impact of Chernobyl is the largest public health problem caused by the accident to date”. Research conducted on the Fukushima-Daiichi accident also reported psychological distress among individuals who felt they had been affected by radiation. Similarly, following the Goiânia accident, more than 112,000 people sought assistance related to symptoms of radioactive exposure; however, only 260 people appear to have been actually contaminated, with four deaths reported.

Counterproliferation

Since the Second World War Allied attacks on the Vemork heavy water production plant, military strikes – and more covert means – have been used repeatedly to target nuclear programmes that are believed to pose a risk of weaponisation. Counterproliferation has, historically, been a leading driver behind the use of military force (or its serious consideration) against nuclear infrastructure. Other instances include: repeated strikes on Iraq’s nuclear facilities – by Iran in 1980, Israel in 1981, and the US in the 1990s; and Israel’s strike on Syria’s Al-Kibar reactor in 2007. As mentioned earlier, there are also multiple instances in which strikes were considered but not carried out.

Logic of Counterproliferation Attacks

The definition of operational success in the context of military strikes as a counterproliferation tool varies on a case-by-case basis and may range from a limited roll-back of a nuclear programme to completely disabling (at least in the immediate term) a country’s ability to continue its pursuit of nuclear weapons. In their work on military strikes as a counterproliferation tool, Kreps and Fuhrmann have proposed that attacks on nuclear facilities may have an impact on the host country’s proliferation activity in four ways: by delaying the production of relevant materials and components through the destruction of chokepoint facilities; by forcing a change in the target country’s approach to fissile material production (shifting from the production of plutonium-using reactors and processing facilities to the production of uranium in easier-to-conceal enrichment facilities); by discouraging foreign suppliers from providing materials and assistance to the programme; and by increasing international inspections of the target country’s nuclear sites.

Military strikes on facilities can also serve a signalling purpose – demonstrating the attacking state’s resolve to do what is necessary to prevent the emergence of a nuclear weapons capability. Israeli strikes on nuclear facilities in the Middle East are a good example of this dual effect. Israeli military action against nuclear facilities has had the effect of physically eliminating or degrading programmes perceived to be a proliferation threat while also reinforcing the so-called “Begin Doctrine” – Israel’s policy of not permitting its regional adversaries to acquire WMD and its preparedness to use “all the means at [its] disposal” to this end. An Israeli nuclear expert interviewed for this paper noted that military attacks on Iranian nuclear facilities may become necessary, not just to physically degrade the programme, which can be reconstituted with some time, but to signal Israeli (and/or US) resolve to continue striking the programme until Iran decides to abandon any nuclear weapons ambitions.

In practical terms, the effectiveness of military strikes as a counterproliferation tool is the subject of considerable debate and depends on a variety of factors. The prevalent opinion among counterproliferation experts tends to be that military force is limited in its effectiveness as a counterproliferation tool and – in isolation – is best-suited to delaying, rather than outright eliminating a state’s nuclear ambitions. However, when accompanied by a large-scale military operation aiming to achieve broader strategic security objectives and/or the installation of permanent IAEA inspectors with the requisite authority to destroy nuclear material – as was the case in the Gulf War and the subsequent approach to dealing with the threat of a nuclear Iraq – it may be able to achieve more lasting results.

Israel’s 1981 attack on the Osirak reactor and 2007 attack on the Al-Kibar reactor are often lauded as examples of successful counterproliferation strikes. Yet, this is a misguided reading of the facts. After the Osirak attack, the Saddam Hussein government only doubled down on its nuclear ambitions; as Ramberg has pointed out in his work on the use of military force as a counterproliferation tool, “Baghdad, which had committed 400 scientists and $400 million to the nuclear program before the attack, enlarged its nuclear staff to 7,000 and upped its budget to $10 billion”.

Following the attack, Iraq also reportedly turned to the Soviet Union for guidance on how to make nuclear facilities more resilient to attack: “Specifically, the KGB taught them how to disguise and disperse industrial facilities, using techniques such as blocking heat sources from being detected by strategic reconnaissance. They also taught the Iraqis how to build in ‘blow-away’ walls to new structures, such that if the building is hit the structure is left sufficiently in place to permit rapid reconstruction.” The Israeli strike on the reactor followed an earlier Iranian attack on the site in 1980, and it was not until US attacks on Iraq’s nuclear sites during the wider Gulf War air campaign that the programme ultimately ended. While the 2007 Israeli attack on Al-Kibar did seem to eliminate Syria’s nuclear proliferation efforts, some accounts attest that Damascus may have considered reconstructing the reactor after the attack.

Israel also did not stop at the 2007 military attack and appears to have continued its use of lethal force as a counterproliferation strategy: in 2008, its special forces assassinated General Mohammed Suleiman who had been in charge of the construction of the Al-Kibar reactor.

Yet, despite these limitations, states have repeatedly resorted to – or seriously considered – military force for counterproliferation objectives. In a 2010 journal article, Fuhrmann and Kreps conducted a statistical analysis of counterproliferation attacks and serious considerations of attack to determine the factors that increase the likelihood of states resorting to military force as a counterproliferation strategy. They found that threat perception in relation to the nuclear programme in question, above all else, drives states’ decision-making on whether to carry out a counterproliferation strike. The authors concluded that decisions to use military force for counterproliferation aims are driven primarily by a fear on the part of the attacking state that the proliferating state “might use nuclear weapons or engage in other offensive behaviour”, not diplomatic recrimination or even retaliatory strikes by the latter. Fuhrmann and Kreps identify three factors that appear to have a particular influence on threat perceptions: the prior existence of militarised conflict; the presence of an autocratic proliferator; and divergent policy interests between the attacking state and the proliferator. Their analysis finds no statistically significant correlation between a decision to carry out military strikes and risks of reprisal or diplomatic recrimination. The authors argue that “states are willing to accept substantial costs in attacking if they believe that a particular country’s acquisition of nuclear weapons poses a significant threat to their security”. Israel’s experience is instructive in this instance; despite considerable concerns among the Israeli decision-makers over the diplomatic backlash the country may have faced in response to its decision to strike Osirak in 1981, the attack went ahead. The decision to resort to military action in the face of possible diplomatic recrimination was informed by a perception among Israeli leadership that there was no other way to prevent an Iraqi nuclear weapon; diplomatic efforts with Iraq had failed and the reactor was about to be loaded with fuel, which would have greatly increased the risk of collateral damage.

Likely Targets and Effects

Depending on the objective of a counterproliferation attack and the nature and state of the programme, a range of facilities and assets may be targeted.

If the purpose of a strike is to cause material damage to a state’s capability to develop nuclear weapons (and not just signal resolve), in instances where a country may be pursuing a uranium weapon, the front end of the fuel cycle is likely to be of greatest relevance for an attacking state. Targets may include uranium mines and mills, conversion facilities, enrichment plants, enriched uranium storage sites and centrifuge production plants. In cases where a state is suspected of pursuing a plutonium weapon, it is more likely to be relying on heavy water reactors, graphite-moderated reactors or fast breeder reactors to accumulate plutonium. These may be dedicated plutonium production reactors or may double as research reactors or power-generating reactors. Light water reactors can also be used for plutonium production, although they tend to be less efficient, as light water reactor fuel tends to contain relatively high levels of Plutonium-240, making it less suitable – although still useable – for the production of nuclear weapons. A reprocessing facility is also required to extract the plutonium from the irradiated fuel. Counterproliferation attacks against plutonium programmes are therefore likely to target reactors or reprocessing facilities – which can pose a serious radiation hazard if material is released. A country may of course host facilities that allow it to pursue either pathway; for instance, Iraq pursued both the uranium and plutonium pathways following Israel’s 1981 attack on Osirak.

Any sites that may be hosting weaponisation-related activities and expertise – such as research reactors and centres or various laboratories – may also be targeted, regardless of the weaponisation pathway a country may be pursuing. Such activities can often be obfuscated in small and non-descript facilities or at sites engaged in civilian or military research. As such, they may be much more difficult to identify for targeting than large enrichment facilities or reactors. Furthermore, weaponisation research sites and even enrichment facilities and centrifuge production plants can be placed deep underground, making them much harder to damage or destroy through the use of limited military force. Assassinations of nuclear scientists have also been a counterproliferation tactic deployed by Israel, as a way of eliminating proliferation-relevant expertise in a state. However, such activity falls outside the definition of “military attack” in this paper.

Energy Disruption

Military strikes on energy infrastructure are an established aspect of waging war. Nuclear power plants thus risk becoming military targets when the disruption of energy production is an objective of a military operation. This is likely to have been a driver behind the Russian occupation of Ukraine’s ZNPP. No other cases examined by the author point to energy disruption as having been a key driver of a military attack on a nuclear facility. However, the case of the ZNPP is unlikely to remain an exception. As nuclear power is expected to play a greater role in the international energy mix, nuclear energy facilities may be targeted for the purpose of energy disruption in future conflicts.

It is worth highlighting that the legality of attacks on energy infrastructure under international law is highly context-dependent. Energy infrastructure is, in many cases, a legitimate military target. However, as discussed earlier, attacks that cause disproportionate harm to civilians or are aimed primarily at terrorising the civilian population are prohibited under IHL. The potential for radioactive release means that attacks which risk the safety of nuclear power plants are particularly difficult to justify under international law since the potential effects of such attacks can lead to significant harm to civilians. However, this does not rule out the possibility that nuclear energy infrastructure and auxiliary systems may be targeted by militaries in the future. As such, it is worth examining the logic of such attacks and possible approaches to risk mitigation.

Logic of Energy Disruption Attacks

One prevalent approach to conceptualising military targeting priorities is underpinned by the idea that strategic objectives can be achieved through the degradation of an adversary’s key physical systems – or centres of gravity (COGs) – as opposed to aimless attrition or a singular focus on degrading an adversary’s military forces. Power generation and transmission infrastructure is a particularly important COG, as its destruction can significantly undermine the adversary’s ability to prosecute a war and incentivise concessions from the political leadership. Energy infrastructure is therefore often an explicit component of military targeting doctrines – including NATO’s Allied Joint Doctrine for Joint Targeting and the US Air Force’s Targeting Doctrine. During the 1999 Kosovo Air Campaign, NATO forces extensively targeted Serbian electrical infrastructure, reportedly destroying 35% of its electricity-generating capacity. The intended effect – ultimately successful – was to wear down the will to resist of the Serbian population and the elites, thus pressuring Serbian President Slobodan Milošević to comply with NATO demands. The US Air Force also targeted Iraqi electrical infrastructure, alongside other strategic targets, during both the Gulf War and the Iraq War. The aim was to incapacitate Iraq’s military while also creating the necessary conditions for the political overthrow of the Iraqi regime. The focus on the degradation of an adversary’s systems, rather than its armed forces, is also central to Chinese military thinking.

Russian military thinking appears to be predicated on similar assumptions, with an apparent recognition of the fact that in modern wars, priorities for the application of force have shifted from the destruction of an adversary’s forces to the destruction of an adversary’s critical facilities. Energy generation infrastructure – including nuclear power plants – is one of the operational-strategic and strategic target categories captured by Russia’s doctrinal concept of Strategic Operation for the Destruction of Critically Important Targets. Strikes on Ukraine’s energy infrastructure have been a key aspect of Russia’s war on the country. Considering that over half of Ukraine’s energy-generating capacity in 2021 came from its four nuclear power plants, it was perhaps not surprising – but still alarming – that these facilities have been targeted. Russian war planning documents for a potential conflict with Japan and South Korea, dating from 2013 and 2014, also included plans to target nuclear power plants, among other energy and CNI targets.

As nuclear power plants tend to have greater baseload generating capacity per station (with the majority falling within the range of 1–4 GW of generating capacity) than other energy sources (although not always), they may be particularly attractive energy-disruption targets: an assault on a single large nuclear plant can cause greater disruption to a country’s energy generation than attacks on smaller hydro, thermal or other generating stations and related transmission infrastructure. In the case of these latter targets, the destruction of multiple targets may be required to produce energy disruption on a similar scale. The capture and shutdown of the ZNPP has had the single most important impact on the loss of Ukraine’s energy production capacity from the incapacitation of a single facility. Prior to the start of the full-scale invasion, the ZNPP had a total net generating capacity of 5.7 GW. The ZNPP is, of course, an outlier as it is the largest nuclear power plant in Europe. Still, its capacity is significantly higher than that of Ukraine’s second-largest power station, the Vuhlehirska Thermal Power Plant, a coal-fired facility with a generating capacity of 3.6 GW, which Russia captured in July 2022. According to some accounts, by summer 2024, Ukraine had lost 25.5 GW of generation capacity. Based on these figures, the loss of the ZNPP, which took the Russian forces less than a day, represented nearly a quarter of all generation capacity lost during the first two-and-a-half years of full-scale war.

Some coal or hydro power plants can rival the generating capacity of nuclear power plants. In such instances, militaries may opt to target these facilities rather than nuclear power generating stations to reduce enemy energy generation capacity and potential collateral effects – namely, the risk of radioactive release. However, the dangers to the population of the destruction of a hydroelectric facility may similarly have devastating consequences if it involves damage to a hydro dam. The breach of Ukraine’s Nova Kakhovka dam caused 58 confirmed fatalities – the actual death toll was likely higher, potentially in the hundreds, by some estimates. The flood affected 620 km2 of land, forced the reduction of energy output from hydroelectric stations along the Dnipro River, destroyed irrigation systems across the region, caused the leakage and spread of chemicals and other pollutants into waterways and soil, and displaced mines, causing a further hazard to human life. As mentioned earlier, attacks on hydro facilities and other power generating plants may, however, not carry the same psychological effect on the population and decision-makers as attacks – or threats of attack – on nuclear infrastructure.

A high dependence on nuclear power for energy production naturally makes it more likely that a country’s nuclear infrastructure will come under threat should it become the target of a military campaign by an adversary. Few countries are currently as dependent on nuclear power as Ukraine. As of 2023, only France and Slovakia generated over half of their energy from nuclear power, with Hungary just below the 50% mark. Another five countries – Finland, Belgium, Bulgaria, Czechia and Slovenia – relied on nuclear power for over a third of their energy generation in 2023. Should an adversary decide to incapacitate energy infrastructure in these countries in the course of a full-scale invasion or strategic air campaign, it is likely that their nuclear power plants or related transmission networks would be at risk. A military assault of that scale remains highly improbable in the aforementioned countries, even though there is concern in a number of them about the threat posed by Russia.

The potential for nuclear energy infrastructure to be targeted in countries that face a clear and considerable threat of strategic military attack from regional adversaries is a greater concern. Countries that may face such threats include Armenia (which has one reactor, making up 31.1% of the country’s energy generation) and South Korea (26 reactors, making up 30.7% of the country’s energy generation). As more countries become interested in nuclear power as a sustainable and clean source of energy, the share of energy produced by nuclear power plants is projected to increase in the coming decade, including in countries that might face strategic-level military threats from adversaries – namely, Iran and South Korea. Concerns have also been raised over the potential for China to attack or otherwise disrupt the operations of Taiwan’s nuclear reactor at Maanshan, should open conflict break out. Taiwan’s ruling party has committed to shutting down the island’s last operating reactor in 2025, which will minimise both the energy security and nuclear safety implications of any future attack. However, Taiwan’s stores of spent fuel withdrawn from the decommissioned reactors will continue to pose a potential radioactive threat should China decide to target them.

Likely Targets and Effects

For the purposes of energy disruption, nuclear installations that may be targeted include nuclear reactors and their support systems, as well as the transmission lines and substations that connect the reactors to the broader grid. The destruction of nuclear fuel production facilities, fuel storage facilities and other sites outside the operational stage of the nuclear fuel cycle – that is, the point in the cycle where energy is being produced – may complicate the operation of nuclear power plants in the long run. However, reactors are normally loaded with fuel that can supply power for years before needing to be replaced. Equally, nuclear fuel can be stored on-site – in cooling pools within reactor buildings – for several years before a shortage of space for spent fuel storage becomes enough of an issue that it impacts the operation of a reactor. Spent fuel pools may also be allowed to exceed their designed capacity if needed. As such, while disrupting the operation of these other facilities may have an impact on nuclear energy production in the long term, an immediate effect on energy generation would require attacks on reactors themselves, their support systems or their transmission infrastructure.

Direct attacks on a reactor vessel are unlikely, if the sole purpose of the attack is to disrupt energy production. The risk of large-scale radiological release if the containment structure and core of a reactor are penetrated – for instance, by a missile strike – is exceptionally high. Instead, militaries aiming to disrupt nuclear energy production or distribution are more likely to target supporting systems and transmission infrastructure. This may be perceived by militaries and governments as a safer and easier alternative to achieving energy disruption objectives – posing fewer risks of a catastrophic nuclear safety incident, eliciting less fervent recriminations from the international community and requiring less firepower to inflict damage. This logic may have been playing out in Russian military assaults on Ukraine’s nuclear infrastructure, with strikes carried out primarily against supporting infrastructure – including water supply, power lines and substations connected to Ukraine’s nuclear power plants, as well as auxiliary buildings and infrastructure at the ZNPP. Ukraine’s reactors have, however, also been hit on a number of occasions – including direct drone strikes on one of the ZNPP’s reactor domes and a drone strike on the New Safe Confinement Structure at Chornobyl. Similar logic has been observed in campaigns targeting non-nuclear energy infrastructure. After receiving significant criticism for attacks on Iraqi energy infrastructure during the Gulf War, coalition forces reportedly limited their attacks during the Iraq War only to energy transmission infrastructure, rather than energy generation stations, as well as switching to weapons expected to be less harmful to civilians. The reduced firepower required to damage power lines and water reservoirs as opposed to reactor vessels (which would require the use of heavy and targeted munitions to penetrate the containment structure) also allows the attacker some credible deniability of responsibility, should diplomatic recrimination be a concern.

However, it is only partially true that attacks on supporting and transmission infrastructure are lower risk: these systems are all critical for the safe operation of a plant. A loss of off-site power (LOOP) incident, a loss of coolant accident (LOCA) or a station blackout (SBO) (when the station loses access to both off-site power and emergency generators) can have catastrophic results for a nuclear plant. This is something that the IAEA has repeatedly drawn attention to in the context of the war in Ukraine. The IAEA has highlighted the risk to nuclear safety posed by the frequent disconnection of Ukraine’s nuclear power plants, and the ZNPP in particular, from off-site power supply and the risks of potential coolant loss following the draining of the Kakhovka reservoir. Off-site power and water supply are critical for the maintenance of a plant’s safe and secure operation. While nuclear power plants have on-site emergency diesel generators and alternative sources of coolant can be sought (for instance, through the drilling of wells), these cannot fully replace normal operating systems or be relied on indefinitely and may themselves be challenging to procure and maintain during a war.

Attacks on transmission networks rather than the generating facilities themselves may also be the preference when a military operation is aimed at the occupation of a country, rather than a purely punitive or coercive military campaign, when destruction rather than territorial gain is the primary objective. In such cases, the adversary may wish to minimise damage to CNI (to the extent that military operational necessity allows) and to prioritise instead targets that may be easier to rebuild following the end of hostilities. In the case of energy infrastructure, transmission lines and substations are much less costly, timely or complex to repair than generating stations. Damaging energy transmission – rather than generation – infrastructure also allows for the rerouting of energy infrastructure away from unoccupied territories and towards those that the adversary has managed to capture.

Disruption of energy production and distribution also has serious implications for both the safety and morale of civilian populations, and potentially also for military operations. Modern societies are highly dependent on reliable electricity supply to operate CNI and critical services – from the provision of water supply, sanitation services and heating to the operation of communication networks, health services, financial systems and government bureaucracies. Disruption of this infrastructure may lead to civilian deaths, at worst, and a decline in morale and willingness to continue with the war effort, at best. For instance, as mentioned earlier, NATO attacks on Serbia’s energy infrastructure elicited frustration among Serbian elites and the broader population that helped pressure Milošević into concessions. These disruptions to daily life may also drive internal displacement and refugee flows outside the targeted country, putting pressure on neighbouring states and potentially decreasing whatever political or military support these states are lending to the targeted country. The potential for renewed waves of refugee flows from Ukraine into Europe has been a concern in light of persistent Russian assaults on Ukraine’s energy system and Russia’s track record of weaponising refugee migration.

Area Denial

Release of radioactive material may be an unintended consequence of a direct attack on a nuclear facility or of other military operations that result in a nuclear accident. However, it is also conceivable that militaries may target nuclear facilities – or threaten to do so – with the express intention of releasing radioactive material precisely because of the risk that such a release would cause to the surrounding environment, as a method of area denial. The author has not identified any instances of nuclear facilities having been expressly targeted for this purpose in the past. Any such attack would also almost certainly violate provisions of the Geneva Conventions and IHL principles more broadly, as it would be difficult to ensure or convincingly make the case that release of radioactive material as a result of attack would not place civilian populations at risk. Furthermore, nuclear facilities in the back end of the nuclear fuel cycle – for instance, spent fuel storage facilities – would have little military utility; attacks on such installations would be difficult to justify on the basis of military necessity. However, a number of experts interviewed for this paper agreed that nuclear sites could hypothetically be targeted by militaries for area denial purposes; as such, the possibility deserves analytical consideration.

Logic of Area Denial Attacks

Attacks to release radioactive material may be used either as part of an offensive operation – denying manoeuvrability and incapacitating an adversary – or as part of a defensive effort – placing radiation in the way of an advancing military. Radioactive or chemical contamination of a territory complicates and delays an adversary’s operations in the affected area and may lead to mission failure. Contamination can have an impact on a force’s freedom of movement, force personnel to operate in cumbersome protective gear, interrupt logistics and supply lines, and require the updating of risk assessments and the adjustment of mission plans. If radiation or chemical exposure is suffered by troops, this may also incapacitate personnel and impact the morale of those not directly affected. Levels of awareness of CBRN risks and how to manage them vary among and within militaries. Furthermore, specialised counter-CBRN (C-CBRN) units ultimately have limited personnel and capacity and may therefore struggle to provide the levels of support necessary to carry out large-scale or complex operations in a contaminated environment or one that poses the risk of a CBRN incident.

Depending on the extent of the radiological or chemical dispersal, a broader emergency response may be needed to address risks to the broader population, thereby drawing on already-stretched civil and military resources. Meanwhile, forces of the attacking state which were previously tasked with holding the now-contaminated territory could probably be reallocated to bolster operations elsewhere on the frontline. Equally, stressing and even exaggerating risks to a nuclear site may be used as a way of deterring any attempts at contesting the territory near a nuclear facility or the nuclear site itself – both because of the actual risk of causing a nuclear accident and because of the diplomatic and public outcry that would accompany such a military operation. Militaries may therefore be incentivised to use these sites as defensive positions. The deterrent and coercive impacts of such threats are discussed in the next section of this chapter.

Despite these challenges of operating in a CBRN environment, there is some reason to doubt the ultimate operational impact and effectiveness of radioactive release as an area denial tool. Most modern militaries have procedures in place to respond to CBRN incidents and to maintain – to the extent possible, based on an assessment of risk to personnel and mission importance – operability in such environments. While radioactive contamination of a territory may delay and complicate a military operation, the adoption of protective measures and – depending on the importance of the mission – some acceptance of risk will likely allow operational objectives to be pursued, albeit with significantly greater difficulty and risk. The measures taken to minimise exposure are dictated by individual militaries’ doctrines for CBRN operations, the nature of the hazard, the availability of protective measures and operational requirements. Protective measures for operations in a contaminated environment include some combination of minimising exposure time, maintaining distance from the hazardous material and employing shielding materials. Trade-offs between the various protective measures may be necessary depending on operational requirements – for instance, foregoing protective gear to minimise exposure time. The degree of acceptable exposure is also dictated by how critical the mission is to achieving operational or strategic objectives.

The unpredictability of any radiological release and the impossibility of controlling the extent of contamination also make radiological dispersal a very risky area denial tactic – one which is likely to pose greater risk than benefit. Radiological dispersal does not discriminate between adversary and friendly forces: any territory denied to the adversary will be equally denied to friendly forces. This not only will complicate friendly force operations in the contaminated territory but – depending on the extent and nature of the contamination – can also make the territory uninhabitable following the end of hostilities. Depending on weather patterns and the nature and direction of the radioactive plume, the territories and civilian populations of the attacking state may also come to harm. The risk of contaminating one’s own territory and putting one’s own troops and civilians in harm’s way is highlighted in NATO doctrinal documents on CBRN operations and was raised repeatedly – alongside IHL considerations – by the military and defence experts interviewed for this paper as a major disincentive for the use of military force against or in the vicinity of a nuclear facility. The extent to which self-harm acts as a disincentive would depend on the attacking state’s concern for its own military personnel and civilians.

The likely eventuality that radioactive contamination spreads to third states would also risk escalating the conflict by potentially invoking a response from countries not already directly party to it. In June 2023, amid rising concerns over the safety of the ZNPP, a bipartisan resolution was introduced in the US Senate calling for any attack on a nuclear facility that disperses radioactive material to NATO territory to trigger “an immediate response, including the implementation of article V of the North Atlantic Treaty”. The resolution was never adopted and the credibility of threats to invoke Article 5 in such an instance are debatable; however, it does highlight the escalatory threat that may be posed by such an incident.

An imperfect comparison can be drawn here with the long history of releasing other “dangerous forces” for the purposes of area denial – for example, flooding territory by breaching dams and dykes. Examples of this practice include the piercing of dykes by the Dutch to break the sieges of Alkmaar (1573) and Leiden (1574) during the Eighty Years’ War; China’s breach of the Yellow River Dam (1938) to slow the advance of the Japanese army during Japan’s invasion of China; the Soviet destruction of the Ukrainian DniproHES hydroelectric dam with the aim of slowing German advances during the Second World War; strikes on the Islamic State-controlled Tabqa Dam by US Special Forces, apparently for the purpose of “territory denial” and to destroy Islamic State garrisons in the walls of the dam; and most recently, Russia’s destruction of Ukraine’s Nova Kakhovka dam in June 2023.

There are of course notable differences between the risks posed by mass flooding and those posed by the release of radioactive material – namely, the much more persistent nature of radioactive contamination, the greater challenges of shielding oneself from radiation, the greater unpredictability of the direction and extent of a radioactive plume and, perhaps most critically, the psychological impact of radioactive contamination. However, as with radioactive release, mass flooding can also cause significant harm to civilian populations and have catastrophic and long-term economic, ecological and health effects – damaging farmland, destroying livelihoods, polluting groundwater and spreading disease. While past instances of mass flooding for area denial have produced some effects on the battlefield, there is a dearth of examples where its use has proven strategically decisive in a military conflict. In some cases, such tactics instead caused adverse effects for friendly forces. The ability of militaries to adapt, especially in pursuit of existential or other strategically important objectives, and the difficulty of controlling these “dangerous forces” makes them a suboptimal method of area denial. As has been noted elsewhere, conventional means – for instance, mines – can be used instead of CBRN or other dangerous forces to achieve area denial with much lower risks to one’s own forces, territory and populations and with lower risks of political castigation.

Likely Targets and Effects

For a substance to act as an effective area denial tool, it must be at once hazardous enough to impede safe operation in the contaminated area and persistent enough to stick around long enough to achieve the desired effect. In this way, the considerations on the use of radioactive material for area denial are similar to those for the use of chemical weapons for that purpose. The definitions of “hazardous enough” and “persistent enough” depend on the intended operational or strategic effects, the intended target (personnel or vehicles) and the resilience of the adversary – including shielding capabilities.

Depending on these considerations, nuclear facilities across the nuclear fuel cycle that contain nuclear materials could in future be targeted for area denial purposes. Broadly speaking, the radiological hazard posed by attacks on facilities in the operational and back end of the nuclear fuel cycle may pose a greater shielding and decontamination challenge than the predominantly chemical hazard posed by materials released from sites in the front end of the nuclear fuel cycle. However, as discussed earlier, the actual risk depends on many factors, which will dictate the type of shielding that personnel wishing to operate in the affected area would require, as well as the amount of time and types of activities that personnel would be able to perform. For an extensive discussion of the physical and psychological effects of nuclear materials release, see the start of this chapter.

Escalation of Violence, Deterrence and Coercion

Besides the potential physical effects of assaults on nuclear facilities, and any consequent release of hazardous material, the threat of such attacks can also have an important psychological impact on populations and political leadership. The fear that radioactive release engenders in the popular psyche – discussed in greater detail at the start of this chapter – makes the psychological dimension conducive to weaponisation, particularly when public opinion and morale, or diplomatic pressure, are important for achieving a given objective.

The combination of the physical, psychological and normative effects of attacks on nuclear facilities means that states may conceive of them as a “half-step” in escalation dynamics between conventional war and the use of nuclear weapons by a nuclear weapons state. This is despite the fact that the effects of nuclear accidents differ considerably from those of nuclear weapons. The conflation of the effects of major nuclear accidents and nuclear weapons detonations sometimes observed in public debate can exacerbate the strategic impact of such threats. Such strikes and threats could be used to signal one’s willingness to escalate a conflict by inflicting – or threatening to inflict – radiological effects on an adversary’s troops, environment and population, without actually resorting to the use of nuclear weapons. The fear of radiological release also means that threats of attacks on nuclear facilities can be used for deterrent and coercive purposes. Threats to the safety of a nuclear site – whether explicit or implied – may be used to deter further escalation by the adversary elsewhere on the front line or away from the battlefield, as a source of leverage in diplomatic negotiations.

Equally, non-nuclear weapons states may strike – or threaten to strike – the nuclear facilities of an adversary in the absence of a nuclear weapons capability for similar signalling and escalatory logic. Non-nuclear weapons states facing a nuclear-armed adversary may use such threats against the latter’s nuclear sites in an effort to compensate for an imbalance in deterrent capabilities. The ability and willingness to threaten to attack an adversary’s nuclear sites may have a significant effect on the deterrent balance between two non-nuclear weapons states and may help to address imbalances in conventional capabilities. One such attempt by a non-nuclear weapon state to bolster its deterrent capability against another non-nuclear weapon state may have played out during the escalation in violence between Armenia and Azerbaijan in 2020. In a statement from its defence ministry, Baku warned that Yerevan “should keep in mind that our armed forces have advanced missile systems in service, capable of conducting high-precision strikes on the Metsamor nuclear power plant, which may result in a huge disaster for Armenia”. Media commentary on the incident suggested that the threat may have been issued following a miscommunication between the two sides and highlighted the escalatory risk posed by such rhetoric.

The possibility of placing adversaries’ nuclear power plants at risk – as a sort of “less apocalyptic” alternative to nuclear weapons deterrence – was raised in public discourse as early as 1978, in an article for Foreign Policy by Chester L Cooper, who was, at the time, assistant director to Oak Ridge Laboratory’s Institute for Energy Analysis. Cooper noted that:

A high explosive bomb, when used against a nuclear target, would acquire some of the radioactive attributes of a nuclear bomb (although it obviously would not have the same blast effects). Thus the presence of nuclear power plants in a given country could effectively transform its enemy’s arsenal of conventional bombs into weapons with radioactive fallout.

He also pointed out that, while some attacks – for instance, those targeting only the electrical system of a nuclear power plant – would likely result in limited consequences that could be mitigated, “no national leader could be confident that an attack would not succeed in destroying the sensitive elements of a nuclear plant”, with catastrophic results. He concluded not only that the proliferation of nuclear power plants would lead to “a vast increase in energy supply for developing nations”, but also that nuclear power generating stations could actually improve relations among countries by introducing a dynamic akin to the concept of mutually assured destruction. Ramberg later built on Cooper’s observations in Nuclear Power Plants as Weapons for the Enemy: An Unrecognised Military Peril, noting that states could threaten to target an adversary’s nuclear facilities in response to being attacked. Ramberg also suggests that nuclear facilities – namely, spent fuel repositories – could be placed along a likely invasion route as part of a country’s defence strategy.

However, as Ramberg also points out, the parallels of the deterrent, coercive or military effects of nuclear weapons and attacks on nuclear facilities are ultimately limited. The effects of nuclear accidents differ in a number of meaningful ways from those nuclear weapons detonations. For instance, even large nuclear accidents – such as the 1986 Chornobyl disaster – do not produce the fireball and blast waves that cause much of the immediate destruction following a ground burst nuclear weapon detonation. Nuclear weapons detonations also produce an electromagnetic pulse, which can disrupt and damage electrical equipment; this is not present during a nuclear accident. Comparisons between the radiation released from nuclear weapons detonations and nuclear accidents (both immediate radiation and fallout) are more difficult to draw, as in both instances the nature, extent and effects of radiation depend on a range of factors.

Nuclear weapons and attacks on nuclear facilities also differ significantly in terms of their operational and strategic utility and significance. Nuclear weapons play a range of roles in countries’ nuclear doctrines and national security and defence planning, most of which cannot be replicated by radioactive release from a nuclear facility. For one, nuclear facilities are, generally speaking, immovable and therefore cannot be used to hold at risk a range of military, territorial or other targets of an adversary. This is further complicated by the difficulties, discussed above, in determining the scale and direction of any radioactive release from a nuclear facility. The immovability of many nuclear installations also makes them less effective for signalling – which a state may do through the deployment or removal of nuclear-armed delivery systems across its territory. As mentioned earlier, some signalling may nevertheless be possible through attacks on nuclear sites – for instance, the use of limited attacks on a suspected nuclear weapons programme to signal resolve to take military measures against the perceived threat, or an attack resulting in radioactive release to signal a willingness to escalate the conflict outside the strictly conventional realm.

Critically, just as attacks on nuclear infrastructure present unique psychological challenges and cross certain normative thresholds, so does the use of even low-yield nuclear weapons. Some have argued that, as a result of the non-use of nuclear weapons since 1945 and the devastation that nuclear weapons are understood to reap on society, a taboo has been established on nuclear weapons use. The breaking of this taboo and of the 80 years of non-use would cross a major normative and psychological threshold. Most of the experts that were asked about the existence of a similar taboo on attacks on nuclear facilities felt that a certain level of restraint, or even a conditional taboo on attacks that resulted in radiological release, does exist. However, they did not go as far as to suggest that the normative restraint on attacks on nuclear facilities reached that of nuclear weapons use. One expert that was interviewed also noted that for countries such as Russia – whose population is willing to accept considerable suffering for the pursuit of the state’s strategic objectives – the radiological threat of an attack on a nuclear facility would likely have limited deterrent effect. In fact, it may produce an inverse “rally around the flag” response.

Nuclear Facilities on the Axis of Advance

So far, this paper has discussed some of the operational and strategic logic that may inform a decision to expressly target nuclear infrastructure. Yet, most militaries, in most circumstances, would not actively pursue such a target – both because of the radiological risks (and associated challenges) that such attacks pose and because of the legal prohibitions and broader normative restrictions on attacking nuclear sites. A number of military and defence experts interviewed for this paper, when asked to consider why a military may be asked to target a piece of nuclear infrastructure, stressed that they saw no justification or operational utility in deliberately targeting a nuclear object.

However, some of the military and defence experts consulted did assess as feasible a scenario in which a military may need to contest control of a nuclear site – or operate in its vicinity – if such an installation were to be encountered on a route of advance during a wider military operation. For instance, the CEZ, which was occupied by Russian forces in February 2022 when they crossed into Ukraine from Belarus, appears to have come under occupation primarily because it was on the route of attack for the Russian Army on its way to Kyiv. The site offers no benefit as an energy target and – relative to the ZNPP – would not have been the most effective target for the purposes of radioactive release considering its close proximity to the Belarusian border. The case of the CEZ is certainly not the first time that a nuclear installation has found itself on a military’s axis of advance. During the Vietnam war, a nuclear research reactor in Dalat ended up in contested territory and at potential risk from military hostilities, despite there being no indication that the facility was expressly targeted. The threat posed to Slovenia’s Krško nuclear power plant during the Yugoslav Army’s invasion in 1991 also does not seem to have been the result of express targeting of the facility (although overflights of the facility at low altitude by fighter aircraft were perceived as an intentional threat to the plant in subsequent accounts). As the number of nuclear power plants around the world is expected to expand in the coming decades, it will become increasingly likely that militaries will have to reluctantly contest or occupy nuclear facilities – be they power plants, fuel production facilities, or spent fuel storage sites.

Concerns over causing a nuclear accident is likely to constrain, complicate and slow a military’s operations in the vicinity of the facility. Should radioactive release occur, those challenges will of course multiply, as discussed in the section on radioactive release for area denial. As described earlier, most modern militaries have access to C-CBRN capabilities with the requisite expertise for conducting operations around facilities that may pose a hazard. However, conducting operations around a nuclear facility – or even occupying a nuclear site – is likely to require a greater number of specialised personnel than what may be available in most militaries, probably requiring operations to be predominantly carried out by unspecialised forces with some C-CBRN support. Furthermore, as already noted, the level of C-CBRN expertise, preparedness and capacity is likely to vary from country to country and may be quite low within some militaries – especially among non-specialist personnel. For instance, the fact that Russian soldiers dug trenches in contaminated soil in the CEZ during their occupation of the site suggests a certain lack of understanding within the ranks of the Russian military of best practices and standards of operation in a CBRN environment. Military activities may also pose an inadvertent risk to nuclear facilities if they include strikes – intentional or accidental – on other parts of a country’s infrastructure (for instance, the energy grid or water supply). Such strikes may not be intended to disrupt the safe operation of nuclear facilities but may nevertheless create nuclear safety risks.

Occupation of a facility also poses a range of challenges for both military and political leadership. Ensuring the safety of an occupied facility would require extensive resources and expertise on nuclear plant operations and nuclear safety. The occupying state would face the challenge of potentially having to maintain nuclear safety in a conflict zone and all of the difficulties that this entails. Political leadership would also need to determine whether and how to manage the occupied facility under existing nuclear safety and security frameworks and who would be responsible for overseeing the operation of the facility in what may still be an active warzone. However, the risk of causing a nuclear safety incident will not always be sufficient to deter a military from contesting the area around a nuclear facility or occupying it, if a military judges that this can be done safely, if the mission is deemed critical enough, or if the threat of radioactive release is not well understood or is disregarded. This was certainly true for the Russian Army, which advanced on both the CEZ and the ZNPP despite the risks.

II. Reducing the Risks of Military Strikes on Nuclear Facilities

So far, the author has outlined some of the operational and strategic effects that militaries may pursue in attacking – or threatening to attack – nuclear infrastructure. As indicated, states may perceive a range of threats – as well as operational and strategic opportunities – that could lead them to use military force against or in the vicinity of nuclear sites. As noted in the introduction of this paper, the intention here is not to normalise the use of military force against or near nuclear installations, but to make clear the need to realistically assess the probability and nature of these risks and to prepare for dealing with them. This chapter considers a few approaches for reducing the risk of military attacks on nuclear facilities. In doing so, it examines the likely effectiveness of these approaches in responding to the various drivers of military attacks discussed earlier in this paper, as well as in mitigating against some of the likely consequences of attacks.

Risk-reduction measures must account for various trade-offs and resource limitations. Even in the most economically advanced or security-minded states, resources are not unlimited; this is likely to be especially true in the context of a broader military conflict, when already-limited sources must be allocated across a range of defence and security priorities. Furthermore, some preventive and mitigation measures may have unintended consequences that could, in fact, increase other undesirable consequences of attack or make attacks more likely. Additionally, an overly cautious approach to risk management – for instance, foregoing nuclear energy altogether out of a fear of possible military threat – may have other undesirable and unnecessary economic, environmental and political drawbacks. As such, the departure point for any risk-mitigation measures should be a realistic assessment of the precise nature and degree of risk that a given state’s nuclear infrastructure is likely to face from military threats.

Strengthening Existing Prohibitions, Norms and Guidance

Since the start of Russia’s full-scale invasion of Ukraine – and, arguably, before that – an important aspect of the debate on how to reduce military risks to nuclear facilities has focused on whether existing legal prohibitions on attacks are sufficient and whether and how they may be strengthened and expanded. Military and defence experts interviewed for this paper repeatedly noted that Western militaries would not attack nuclear facilities, with IHL considerations and existing restrictions on attacks under the Geneva Conventions repeatedly referenced. This awareness and commitment to existing restrictions is reassuring and suggests that these are likely to be sufficient to prevent attacks from states that are committed to respecting international law. Of course, some states may chose to disregard their IHL obligations or to find ways to justify their actions under IHL – even when such justifications are unconvincing or disingenuous.

However, as the analysis presented in this paper has highlighted, there are instances when military force may be used against, or in the vicinity of, nuclear installations for the pursuit of legitimate military objectives. To ensure that such objectives can be pursued with minimal threat to nuclear safety, to civilians and to the environment, additional clarity and guidance on the operationalisation of existing relevant legal provisions may be useful. These may be developed by individual governments or groups of governments, or by international organisations, such as the International Committee of the Red Cross. Ensuring that these principles and their operationalisation are understood and respected at all ranks of a military force is also critical.

As discussed, a potential scenario in which a military may find itself having to conduct operations around – or directed at – a nuclear facility is one where it encounters a nuclear site on an axis of advance and needs to contest control over or occupy the site. It is unclear to what extent various national nuclear regulators or militaries have thought through their country’s ability to carry out such operations, what procedures may have to be adopted and what stakeholders (military and civil) would need to be engaged to ensure the safe transfer of control of a nuclear site in the course of a military operation. Laying out such operational considerations should be integrated into military and civilian planning well in advance of any military operation that may involve such a scenario. Militaries would also benefit from wargaming possible scenarios in which they may be required to contest territory that hosts a nuclear facility – to develop operational options for doing so safely, assess readiness and identify challenges. For instance, one possible approach may be to quarantine the facility and cut off supply chains to any military units operating in its vicinity. Such an approach, however, must account for the need to continue to allow supplies and personnel to reach the nuclear site itself, so as to ensure its safe operation – a vulnerability which could in turn be weaponised by the military occupying the site.

At the same time, preparing to use force against or near nuclear installations cannot be normalised. Clear demonstrations of political commitments to existing restrictions and norms relating to attacks against nuclear facilities should be pursued to this end. As mentioned earlier, a limited taboo on attacks on nuclear facilities that result in radiological dispersal may exist or be emerging. Any such taboo should be strengthened. This can be done first and foremost by continued non-attack on facilities that may reasonably be expected to result in significant release of radioactive material, and the condemnation of any military action that presents such risks. Individual states may also wish to explicitly reiterate their own commitments to the relevant provisions of the Geneva Conventions in unilateral or multilateral statements.

Some have also argued that further NAAs, modelled on the India–Pakistan NAA, should be pursued. However, the relevance of the NAA as a model for future agreements is likely to be limited, due to the particular historical, political and security context in which it was concluded. The 1988 Agreement was concluded between two countries with advanced nuclear programmes; India had tested its first nuclear device 14 years earlier, while Pakistan was known to be pursuing a nuclear weapons capability at the time. The mutually perceived vulnerability of the two states, exacerbated by rising bilateral and global tensions but which had not yet escalated to armed confrontation, created the conditions for such a risk-reduction measure. A similar NAA would – of course – not be currently possible between Russia and Ukraine. An Iranian nuclear expert interviewed for this project rejected the possibility of a similar NAA in the Middle East, pointing to the perceived imbalance of vulnerability created by Israel’s undeclared nuclear weapons capability. An Israeli interviewee also rejected the possibility, arguing that Israel faces an existential threat from its regional adversaries and noting the historical propensity of regional actors – including Syria, Iraq and now Iran – to obfuscate their apparent pursuit of nuclear weapons. One possibility might be the conclusion of bilateral or multilateral NAAs between states with friendly relations – for instance, Argentina and Brazil. Such agreements might serve as an example-setting and norm-strengthening exercise; however, the practical effects of such agreements would be limited and their conclusion outside of any clear security need to do so may instead engender suspicion between parties.

There are currently very few international standards for the safe and secure management of a nuclear facility in a warzone or under military occupation. The Seven Indispensable Pillars and Five Concrete Principles for Nuclear Safety and Security articulated by the IAEA since the start of the war in Ukraine offer a useful starting point for establishing best practices for nuclear safety and security in warzones. Yet, they ultimately fall short of offering concrete guidance and processes for nuclear safety and security in armed conflict. The technical guidance document on challenges in the application of the IAEA safety standards and nuclear security guidance during armed conflict (under development by the IAEA at the time of writing) is also expected to draw on existing IAEA guidance. The Agency’s consensus-based approach to decision-making, the need to maintain flexibility of response to address the unique challenges and needs of a given facility or attack scenario, and the general aversion of IAEA member states to any overly prescriptive standards on the operation of their nuclear facilities ultimately limit the Agency from going far beyond providing fairly high-level direction and ultimately drawing on existing nuclear safety documents. This should not, however, preclude individual states from ensuring that their nuclear regulators and operators, emergency response services and militaries have the necessary standards and operational plans in place to both respond to the threat of attack on their nuclear facilities (commensurate with the risk profile) and conduct military operations in the vicinity of nuclear sites in other countries, if necessary.

Strengthening Passive and Active Defences

One option for reducing the risks of military strikes against nuclear sites may be to strengthen the physical protections of the facilities in question. Increased passive defences could help to reduce the likelihood of damage to systems and equipment – whether intentional or accidental – while also limiting the spread of toxic material should any be released. Decreasing the degree of damage an attacking state may be able to inflict on a given target (or increasing the costs of inflicting meaningful damage) may also act as a disincentive for attacking forces. This is likely to be particularly true when attacks on other, softer, targets may produce similar effects – for example, in instances of attacks carried out for the purpose of disrupting a country’s energy production.

Most nuclear power plants operating today are already protected by reinforced concrete containment structures, making them challenging, although certainly not impossible, to penetrate. Spent fuel storage normally also benefits from reinforced concrete containment and can withstand substantial force. More broadly, the IAEA requires national regulators to ensure that facilities across the nuclear fuel cycle, including conversion and enrichment sites, benefit from containment structures and other safety measures that are appropriate for a range of assessed risks. Such assessments are normally completed as part of the Design Basis Threat (DBT) process, which requires states to account for such threats as severe weather (including floods and strong winds) as well as terrorist attacks. Threats that fall outside what a nuclear regulator can reasonably be expected to account for are classed as beyond-DBT threats and are generally the responsibility of the state. This would include military attacks and occupation of facilities. Since the full-scale Russian invasion of Ukraine, the need to update IAEA requirements and guides to account for the military threat has been the subject of considerable debate within expert and regulatory communities.

Constructing thicker containment structures over existing reactors that already benefit from reinforced containment would be a resource-intensive undertaking which would be incommensurate with the risk of an accidental strike and ineffective against an adversary determined to penetrate the reactor core to release radioactive material. Options for strengthening protections of spent fuel may include the construction of additional containment structures, moving dry spent fuel casks inside specially designed facilities (in instances where casks are currently stored outside) or moving storage underground. Conversion and enrichment facilities may also be moved underground or under reinforced containment structures if deemed necessary.

In all these cases, additional physical protection would be subject to cost–benefit considerations and would be dependent on the extent and nature of the threat faced by a given state. For instance, it may be reasonable to expect that Ukraine or even Finland would expend the necessary resources to strengthen the containment structures around their wet spent fuel storage facilities to withstand a determined large-scale kinetic strike (if current structures were assessed to be insufficient), considering the Russian threat. Similar measures would seem excessive for Canadian or UK spent fuel storage facilities, which are likely at more significant risk from cyber attack than missile strikes. Beyond-DBT and broader national-level risk assessments must therefore take into account the likelihood of potential military attacks on their nuclear sites by hostile states.

In addition to the nuclear facilities themselves, states may consider strengthening the physical protections around auxiliary systems and equipment that ensure the safe operation of nuclear facilities and which do not normally benefit from similar levels of protection. As discussed earlier in the paper, in case of a military conflict, an adversary is likely to target this auxiliary infrastructure as part of a campaign to disrupt energy production and distribution. This may include infrastructure for the supply of water to the facility (such as water pipes and pumps) and parts of the power grid supplying and offloading electricity to nuclear plants and substations – namely, power lines and substations. However, such an approach is also not a panacea against a determined attacker.

Ukraine’s experience of improving physical protections of electrical substations is instructive. The country has adopted a three-tier approach to the protection of CNI. Varying degrees of protection are offered to facilities depending on the assessed risk profile and size of the potential target, with the use of sandbags, gabions and netting around transformers to protect equipment from UAV attacks, and the construction of T-walls and other cement structures around individual transformers and whole substations to provide defence against missile attack. However, Ukraine’s experience also demonstrates the limits of this approach. The construction of passive defences is likely to be most effective in advance of a war, as such efforts run up against a number of challenges during an active conflict. First, defences – particularly large cement structures – can become costly to construct when resources are likely to already be tight. Resource constraints have reportedly slowed – and, in some instances, halted – Ukrainian construction of larger concrete structures around substations. Second, construction of such defences also takes time and is quite conspicuous, easily detected by an adversary’s reconnaissance. Defences can be destroyed before they are ever completed. Third, adversaries learn to adapt their methods of attack to counter constructed defences. In Ukraine, Russia has reportedly started fuelling its missiles with metal-corroding fuel to penetrate the passive missile defences the Ukrainians built around their substations. Finally, as with the containment structures around nuclear reactors, reinforced concrete structures will only delay and complicate – but not stop – the efforts of a determined adversary.

Finally, the perception that a site is not particularly vulnerable, or that an attack is unlikely to result in significant release of hazardous material, may, in fact, increase the likelihood of it being targeted in some instances. For example, a state may feel that – due to the presence of a reinforced containment structure – it can safely carry out UAV strikes against a nuclear reactor, for instance, to force the facility to shut down and thus stop generating power or to signal resolve to use military force against a suspected proliferation threat. Militaries may also generally be less worried about operating around facilities they perceive to be well-protected, possibly increasing the risk of military force being used near nuclear sites. Such action does, of course, run a high risk of miscalculation and may ultimately result in catastrophic effects.

States may also be more willing to carry out counterproliferation attacks against sites which they do not believe pose a serious risk of toxic material release. For example, in conversation with the author, David Deptula – who was the principal attack planner for Operation Desert Storm during the Gulf War, which included attacks on Iraq’s Al-Tuwaitha complex – noted that considerations around the potential for the spread of radioactive material were a factor in targeting decisions. The operation ultimately went ahead, as the risk was deemed to be low because the facility was surrounded by a large sand berm. An Israeli nuclear expert interviewed for the paper, when asked about whether the risk of radioactive release from Iranian nuclear facilities would be likely to be factored into Israeli decision-making on whether to carry out military strikes against Iranian sites, confirmed that such consequences would likely be considered. However, they also noted that – in the case of strikes on enrichment facilities – the threat would be primarily chemical, not radiological, and the fact that some of the sites are underground and most are away from population centres would mean that the actual threat to civilians would be low – seemingly suggesting that this would likely not act as a major inhibitor to a decision to attack. Similarly, states may be incentivised to increase the potential health or political consequences of an attack – for instance, by introducing highly radioactive material into a facility under threat of attack – as a way of deterring attacks.

For these reasons, any increased passive defence of facilities may be combined with active defence efforts to improve effectiveness against attack. Active defence may include a combination of an increased military ground presence in the vicinity of key sites (either permanent or deployed in response to an active threat) and the deployment of air defence systems. Ground forces must include a meaningful C-CBRN presence and specialised forces trained to operate in complex environments – ideally, in sufficient numbers to not only provide advice to the units that may have to defend the plant but to play a leading role in any defensive operation. The same should be true for any military forces expecting to have to advance on territory where nuclear facilities and material may be present. As has been reiterated throughout this paper, military operations directed at or in the vicinity of nuclear installations should not be normalised; however, militaries need to be prepared to contest legitimate military objectives in the vicinity of nuclear installations in a way that minimises the risk of causing a nuclear safety incident.

Air defence systems are likely to be the most effective active defence measure that a country can take to defend its nuclear sites. They are generally deployed near nuclear facilities by states likely to be at particularly acute risk of strikes on their nuclear infrastructure. Calls for additional Patriot systems and the perception that this is the only way to ensure the protection of the country’s nuclear facilities came up repeatedly in the author’s interviews with nuclear operators and policymakers in Ukraine. However, the provision of effective air defence for nuclear facilities is not a panacea and must consider a number of factors. Air defence systems are costly to field and limited in number, even for well-resourced militaries. As such, other potential targets – such as major population centres, key military assets and other CNI – may be prioritised over nuclear sites for air defence allocation, particularly if large-scale radioactive release from an attacked facility is deemed unlikely. Furthermore, providing point defence for individual facilities is complicated by the fact that a facility might be at risk from a variety of threats, which may include guided or unguided systems and everything from ballistic missiles to low-altitude cruise missiles and UAVs. Each threat requires different air defence capabilities. As such, effective air defence for nuclear facilities needs to be layered and distributed and is therefore likely simply to form part of a country’s broader air defence system. Ultimately, as with containment structures, air defence is not a guaranteed solution for defending against a determined adversary.

Finally, as with increased passive protection, strengthening the active defence of a nuclear facility may risk having a counterproductive effect by increasing the likelihood that the site is targeted or that threats of attack may be used for deterrent or coercive purposes. Placing military personnel or equipment in the vicinity of a nuclear site for defence of the facility may draw fire on those positions, thereby also putting the safety of the facility at risk. States may also argue that the placement of military units or systems near a nuclear site by an adversary is intended to grant protection to the military assets, rather than the nuclear infrastructure – whether or not that is actually the case. The argument might then be made that the facility has lost its protections under the Geneva Conventions. This would be an erroneous reading of international law if an attack on the facility is expected to result in significant harm to civilian populations, yet this would likely not stop a determined attacker from making the argument.

Redundancy and Distribution

Minimising the effects – and, by extension, the incentives – of attacks on nuclear facilities may be achieved by increasing redundancy in systems and distributing potential targets. Increasing the number of targets that need to be destroyed or seriously damaged for an attacking state to achieve a desired effect may help to disincentivise attacks on nuclear sites by tilting the cost–benefit calculus for the attacking state. Not only does the attacking state need to expend greater resources for a given attack, but also smaller facilities can be reconstructed quicker than larger ones and redundant systems can continue operating by relying on parts of the system that escaped attack. Furthermore, the scale of force needed to eliminate a highly redundant and distributed system poses a greater risk of the conflict escalating beyond what may be an acceptable threshold for the attacking state than a single strike on a centralised facility. Such an approach is likely to be particularly effective against military strikes carried out for counterproliferation or energy disruption purposes, when the intention is to have an impact on a programme or system as a whole, rather than to release (or threaten release of) toxic material – which can conceivably be achieved by attacking a single site. The practical challenges and escalatory risks of using military force to meaningfully degrade the Iranian nuclear programme, as well as the system-wide assault on Ukraine’s energy system (see Box 3), demonstrate the defensive value of decentralisation and redundancy well.

Box 3. Decentralisation and Redundancy of Ukraine’s Energy Grid

In response to Russia’s attacks on the country’s energy grid since the start of the full-scale invasion, the need to decentralise Ukraine’s energy system has become abundantly clear and is being prioritised. The use of emergency generators by businesses and private citizens has been one example of decentralisation – an organic adaptation by Ukrainian society to the disruption of the main energy system. But top-down initiatives to speed up decentralisation are also in place. The Ukrainian government is offering incentives to civilians, businesses and housing cooperatives for the installation of small-capacity local power generation. The US has also been working with municipalities across Ukraine to provide small and medium gas-powered cogeneration units.

As the Ukrainian grid is reconstructed, the focus appears to be on the construction of greater numbers of smaller capacity energy generation systems. The greater integration of renewable energy sources such as solar and wind into the Ukrainian grid is also being pursued, both to meet European green energy generation standards and to increase resilience in the grid, which has historically relied heavily on nuclear power and – to a lesser extent – coal, gas and oil. Independent experts have also stressed the benefits of diversification and decentralisation for ensuring the resilience of Ukraine’s energy grid, through the introduction of microgrids, small modular gas turbines and renewable energy sources (solar and wind) as well as the use of batteries for energy storage.

The relatively large generating capacity of nuclear power plants contributes to the centralisation – and therefore decreases redundancy – of an electrical grid. As mentioned earlier in this paper, this may make nuclear plants particularly attractive targets for incapacitation by an adversary aiming to disrupt energy production. Decentralisation in nuclear energy generation could be achieved by eventually phasing out traditional gigawatt-scale nuclear power plants in favour of the deployment of small modular reactors (SMRs). SMRs have generating capacities of less than 300 megawatts and are physically smaller in size. They may therefore be used to establish more decentralised grids while also being easier to build physical protections around or even build underground. SMRs can be designed to operate with accident tolerant fuels and to be passively cooled, meaning they would not require electricity to pump water for fuel cooling (although some non-SMR reactors may also be able to integrate similar features in their designs). However, the transition from traditional gigawatt reactors to SMRs will take time. SMR technology has yet to be deployed and many of today’s operational large reactors are due to continue operation for several decades, with many more being constructed.

Yet, while the decentralisation of nuclear programmes may decrease some risks of military strikes, it may exacerbate others. A greater number of smaller nuclear facilities distributed across a country might provide a larger target set for states that mean to intentionally target nuclear facilities – for instance, for area denial or coercive/deterrent purposes. It may also increase the likelihood of militaries coming into contact with a nuclear facility in the course of an operation, thereby exacerbating risks of unintentional damage to nuclear infrastructure.

Whole-of-Society Resilience

An important part of the effort to defend against, disincentivise and mitigate the consequences of attacks and threats of attack on nuclear facilities should be the strengthening of whole-of-society resilience against such attacks and threats.

Such lines of effort can help to reduce the effects – and, by extension, some of the incentives – of military strikes on nuclear facilities across most of the drivers discussed in this paper. However, they are likely to have the greatest impact on reducing the effectiveness and consequences of attacks aimed at inciting public and political anxiety and/or degrading public and leadership morale – namely, energy disruption and coercive attacks or threats of attack. As discussed earlier, while disruption of nuclear energy generation may in part be intended to undermine military production as one of its objectives, it is more often than not meant to degrade the morale of the civilian population by direct denial of reliable energy and disruption of industrial and economic activity. The use of attacks – or threats of attack – for deterrent and coercive purposes may be primarily targeted at a country’s leadership; yet, the latter is likely to be influenced by public fears and demands (at least in a democracy).

Increasing resilience within the population for dealing with the consequences of attacks on nuclear sites, or nuclear safety incidents more generally, may entail a few different lines of effort. This can include broader dissemination among the general population, local governments and emergency responders of information on dealing with the immediate risks of low-level radiation and chemical hazards (including sheltering in place without forced ventilation and simple decontamination measures such as the removal of contaminated clothing). Communities should also be prepared to deal with widespread and prolonged energy disruption, potentially with limited support from central government if the disruption is widespread across a country. This may include ensuring that communities – as well as individual households and businesses, where logistically feasible – are equipped with independent power generation and water sanitation capabilities. Promoting basic nuclear safety education – among both the public and political decision-makers – might help to combat susceptibility to panic and misinformed responses and improve resilience to the threat. It may also be appropriate to expand the number of medical service providers in a country that are trained to deal with large-scale radiological dispersal, to ensure that medical facilities outside of just specialised centres are able to provide support in case of a large emergency.

As with other risk-mitigation measures, it is important to ensure that public awareness-raising, preparedness and resilience efforts are commensurate with the actual risk faced by a state. Otherwise, one runs the risk of alarmism, inciting unnecessary anxiety among the population. This is likely to result in undesirable consequences, as discussed earlier in this paper. However, as Brooke Rogers has noted, communication of risks to the public need not necessarily result in a panicked public response and fears of the potential for the latter should not preclude efforts to responsibly and thoughtfully engage with the public on this subject. Avoiding alarmism and panic among the population requires the careful formulation of risk-communication strategies. As others have noted elsewhere, risk communication on CBRN threats must prioritise establishing trust between government and the public. This requires the formulation of a clear and unified message, an honest articulation of both what is known andwhat remains unknown, and actionable instructions on what measures to take. Ensuring that external-facing versions of national-level risk assessments – such as the UK’s National Risk Register – are made available and socialised in condensed and easily-digestible formats to the general population will help to socialise the risks at an appropriate level. To ensure effectiveness, communication on potential nuclear safety or energy disruption risks should not be restricted to crisis scenarios. Thoughtfully developed communication strategies should allow governments to sensitise populations to the nature of the risk and to appropriate risk-mitigation strategies at various stages – not just at the time of an acute crisis.

Another key aspect of raising societal resilience should be efforts to counter disinformation and misinformation which may be used to weaponise future nuclear safety incidents. Such efforts are particularly important in response to the rising prevalence of the broader weaponisation of disinformation by countries such as Russia and China. The unintentional spread of misinformation can have equally pernicious effects. For instance, frequent comparisons in media – and even sometimes by political leadership – between possible worst-case scenarios at the ZNPP and the 1986 Chornobyl accident are inaccurate and risk inciting panic among the general public and decision-makers. Such reactions are highly detrimental to sensible policymaking, as they grant Russia outsized leverage to practise extortion by threatening the facility’s safety. Similar disinformation and misinformation tactics might be leveraged to exacerbate the deterrent and coercive impact of threats against nuclear facilities in other contexts – for instance, in a potential future conflict between China and Taiwan.

Conclusion

Russia’s full-scale invasion of Ukraine has brought public and policy discussion on the risks that military activity poses to nuclear installations to the fore. Ukraine’s case is, in many ways, unique. However, nuclear installations have been targets of military attack throughout the history of nuclear weapons and nuclear energy; they are likely to face similar risks in the future. While the use of military force against or in the vicinity of nuclear installations should not be normalised, considering the operational and strategic logic behind such military activity – as well as its likely consequences – can help political and military leadership prepare to defend against related threats and mitigate risks.

This paper has aimed to contribute to ongoing efforts on the topic of nuclear safety in armed conflict by improving understanding of the strategic and operational drivers for the use of military force against and near nuclear installations. These include: counterproliferation; energy disruption; area denial; efforts to generate escalatory, coercive and deterrent effects; and military operations around or at a nuclear facility encountered on the axis of advance. As the number of nuclear installations around the world is expected to grow, militaries, national and international policymakers and regulators need to consider the possibility that militaries may increasingly run the risk of encountering and having to operate or contest control of nuclear sites in the course of military operations. Militaries that are determined to respect the principles of IHL and the safety of civilians and the environment more broadly, but which may find themselves in such situations, should have the necessary guidance and tools to pursue legitimate military objectives safely. At the same time, countries should be prepared to protect and ensure the nuclear safety of nuclear installations that may be targeted by states that do not share a similar respect for IHL.

The paper has put forward a number of considerations and recommendations that may be helpful to informing ongoing efforts within multilateral organisations, among policymakers and in militaries for reducing the risks that military activity may pose to nuclear installations. These include: identifying opportunities to strengthen effective implementation of existing restrictions on attacks on nuclear facilities and reiterating political commitments to respecting these restrictions; considering options for passive and active defence of facilities and auxiliary systems during active conflict, including through the deployment of layered and distributed air defence and specialised C-CBRN ground capabilities; increasing redundancy and decentralisation of nuclear installations and related networks; and improving whole-of-society resilience to nuclear safety incidents, including through efforts to raise levels of emergency preparedness, risk awareness and education on nuclear safety among the public and political leadership.

Each of these risk-mitigation measures, of course, has to compete for limited resources, especially in the instance of an ongoing military conflict. Each also needs to account for unintended consequences, which may, in fact, increase the likelihood and consequences of attacks on nuclear sites in some instances. Thoughtful management of risk and corresponding trade-offs requires an understanding of the motivations for which an adversary may attack a given facility, to appropriately and effectively address the incentives for a potential attack. Considering which consequences of an attack are of greatest concern and the best approaches to mitigate them also helps with the prioritisation of risk-management efforts.

Annex

▲ Table 1: Selected List of Instances of the Use of Military Force Against or in the Vicinity of Nuclear Installations. Note: The author does not purport this to be a comprehensive list of all instances of the use of military force against nuclear facilities. Furthermore, the table does not capture instances where attacks have been seriously considered or threatened but not carried out. For a partial list of such instances, see Matthew Fuhrmann and Sarah E Kreps, “Targeting Nuclear Programs in War and Peace: A Quantitative Empirical Analysis”, Journal of Conflict Resolution (Vol. 54, No. 6, 2010).

Darya Dolzikova is a Senior Research Fellow with RUSI’s Proliferation and Nuclear Policy programme. Her research focuses on the strategic aspects of civil nuclear and nuclear weapons technology. Darya’s particular areas of expertise are on the Iranian nuclear programme and on military threats to nuclear facilities. She has also conducted extensive work on broader nuclear proliferation trends, the use of sanctions as a counter-proliferation tool and Russia’s role in global nuclear supply chains.