Nuclear power produces 0.03 deaths per terawatt-hour of electricity. Coal produces 24.6. Chernobyl — the worst nuclear accident in history — killed approximately 4,000 people total. Fukushima killed zero from radiation. Meanwhile coal kills hundreds of thousands per year through air pollution that most people never connect to an energy source. The fear gap between nuclear's actual risk and its perceived risk is one of the best-documented mismatches in psychology. This investigation examines the data, the psychology, and the real problem with nuclear — which is not safety.
The comparative safety data for energy sources has been compiled by multiple independent research groups and the results are consistent. On deaths per terawatt-hour of electricity generated — including all deaths from mining, plant construction, operation, accidents, and air pollution from combustion — the ranking is unambiguous:
These numbers include the worst nuclear accidents. Chernobyl is in the 0.03 figure. Fukushima is in the 0.03 figure. Three Mile Island is in the 0.03 figure. With everything counted, nuclear power has killed fewer people per unit of energy than any fossil fuel by a factor of hundreds — and roughly equals wind and solar on the same metric.
The 1986 Chernobyl disaster is the worst nuclear power accident in history. The explosion and subsequent fire released radioactive material across a wide area of Ukraine, Belarus, and Russia. The incident is frequently cited as evidence of nuclear power's inherent danger. The evidence is more specific than that.
The UN Scientific Committee on the Effects of Atomic Radiation (UNSCEAR), the authoritative international body on radiation health effects, estimates approximately 4,000 deaths total attributable to Chernobyl — including 28 first responders who died of acute radiation syndrome within months of the accident, and roughly 3,500–4,000 thyroid cancer deaths among the 6,000 children who developed thyroid cancer from radioactive iodine exposure (with a survival rate of approximately 98%). The UNSCEAR figure is the scientific consensus; alternative estimates using linear no-threshold models produce higher numbers (15,000–60,000) but these are contested on methodological grounds by UNSCEAR and most radiation scientists.
The Chernobyl exclusion zone — 2,600 square kilometers — remains officially off-limits to human habitation, though scientists have documented thriving wildlife populations within it. This is the visible legacy of the disaster that shapes public perception. What is less visible: the chronic health toll of coal combustion in the same region and globally, which kills more people per year than Chernobyl killed in total, every year, without a memorable event to anchor the perception.
In March 2011, a magnitude 9.0 earthquake and resulting tsunami struck Japan, disabling cooling systems at the Fukushima Daiichi nuclear plant. Three reactor cores experienced meltdowns. Approximately 154,000 people were evacuated. The event triggered nuclear shutdowns across Japan and Germany and contributed to nuclear phase-out policies in multiple countries.
"The radiation doses received by workers and members of the public from the Fukushima Daiichi accident were, in most cases, low and unlikely to cause observable health effects."
UN UNSCEAR 2020 Report on Fukushima — the definitive international scientific assessmentAccording to the UN UNSCEAR 2020 report — the comprehensive scientific review of Fukushima's health effects — zero deaths from radiation exposure have been attributed to the accident. One cancer death has been classified as potentially attributable to occupational exposure under Japanese government criteria. By contrast, the evacuation itself is estimated to have caused approximately 2,200 deaths — from disrupted medical care for the chronically ill, psychological trauma, and the physical toll of evacuation on elderly and infirm populations.
The scientific evidence is that the radiation released at Fukushima — while real, measurable, and requiring evacuation as a precaution given uncertainty — caused less harm than the evacuation ordered in response to it. This is not a fringe position. It is the conclusion of the UN's own scientific committee.
Why does nuclear power carry disproportionate fear relative to its actual death toll? Psychologist Paul Slovic at the University of Oregon has spent decades researching what he calls the "psychometric paradigm" — the factors that cause humans to overestimate some risks and underestimate others. His findings are directly applicable to nuclear.
Risks that score high on "dread" — perceived lack of control, catastrophic potential (even if low probability), involuntary exposure, unknown mechanisms, and delayed effects — generate disproportionate fear regardless of actual mortality statistics. Nuclear power scores high on nearly every dread dimension: invisible radiation, perceived involuntary exposure, contamination of large areas, long-lasting effects (radioactive half-lives), and the visual association with nuclear weapons. Coal combustion, which kills orders of magnitude more people, scores low on dread: the danger is diffuse, the mechanism is invisible (air pollution), it feels voluntary (you can choose not to drive), and there are no dramatic accident images to anchor perception.
The safety argument for nuclear is strong. The cost argument is not. New nuclear plant construction costs in Western countries have escalated dramatically over the past 50 years. Academic research by Arnulf Grubler (2010) documented a systematic pattern of cost escalation in French nuclear construction — the opposite of the learning curves that have driven solar and wind prices down by 90%+ over the same period.
The Vogtle Plant in Georgia — the first new U.S. nuclear plant in decades, completed in 2023 — cost approximately $35 billion for roughly 2.2 gigawatts of capacity, or about $16,000 per kilowatt. Hinkley Point C in the UK is projected to cost approximately £33 billion for 3.2 GW. For comparison, utility-scale solar currently costs approximately $1,000–1,500 per kilowatt. Natural gas combined cycle plants cost approximately $700–1,000 per kilowatt.
The cost problem is real but not universal. South Korea builds nuclear plants at approximately $3,000–4,000 per kilowatt. China is building at similar costs. The gap between Western and Asian construction costs suggests the problem is partly regulatory complexity, partly one-off custom engineering rather than factory production, and partly the cost of financing during multi-decade construction timelines. Small modular reactors (SMRs) — factory-built, standardized units — are the current industry bet on solving the cost problem. The first commercial SMR projects in Western countries are in early construction phase as of 2025.
France generates approximately 70% of its electricity from nuclear power and has the lowest per-capita carbon dioxide emissions from electricity of any major industrial nation. The French grid produces roughly 60 grams of CO₂ per kilowatt-hour; the U.S. grid produces approximately 400g; Germany's produces approximately 350g. Germany — which accelerated its nuclear phase-out following Fukushima — has seen its electricity-sector emissions remain substantially higher than France's despite massive investment in solar and wind.
The IEA's Net Zero Emissions by 2050 scenario explicitly requires nuclear capacity to approximately double globally by 2050. The scenario is not based on preference — it is based on the grid stability math: solar and wind are variable (dependent on sun and wind), and long-duration energy storage at the scale required to fully replace dispatchable baseload does not yet exist at viable cost. Until it does, dispatchable low-carbon generation — nuclear, hydropower, and geothermal — performs a grid function that renewables alone cannot.