nonacademicresearch.org Editorial
Nuclear power is widely perceived as one of the most dangerous energy sources. By the evidence — deaths per unit of electricity generated — it is one of the safest. Deaths per terawatt-hour of electricity: coal 24.6, oil 18.4, gas 2.8, nuclear 0.07. This inversion of public perception and empirical record has measurable consequences for energy policy, climate outcomes, and public health.
Nuclear power is widely perceived as one of the most dangerous energy sources. By the evidence — deaths per unit of electricity generated — it is one of the safest, comparable to wind and solar and far safer than any fossil fuel. Deaths per terawatt-hour of electricity: coal approximately 24.6, oil 18.4, gas 2.8, nuclear approximately 0.07. This inversion of public perception and empirical record has measurable consequences for energy policy, climate outcomes, and public health, including through the aftermath of Germany's nuclear phase-out following Fukushima.
When the nuclear reactor at Three Mile Island partially melted down in 1979, and when the Chernobyl reactor exploded in 1986, and when three reactors at Fukushima Daiichi failed following the 2011 tsunami, each event produced extensive news coverage, public alarm, and lasting shifts in energy policy. Nuclear power has been the subject of more sustained public fear than any other energy technology.
This fear has consequences. Germany phased out its remaining nuclear plants following Fukushima, replacing much of the lost capacity with coal. Multiple countries have cancelled planned nuclear capacity expansions in response to public opposition. In many jurisdictions, nuclear power requires more regulatory compliance per unit of power produced than any other energy source — reflecting fear more than empirical risk.
The question of whether this fear is calibrated to the evidence has a clear answer in the academic literature: it is not. Nuclear power's mortality record, comprehensively examined, places it among the safest electricity sources in the world by the measure that should matter most for energy policy decisions — deaths per unit of energy produced.
The most systematic compilation of energy-related mortality data comes from Benjamin Sovacool and colleagues, whose work in Energy Research & Social Science (2016) examined 216 energy accidents from 1907 to 2007, augmented by routine pollution mortality estimates. Their estimates of deaths per terawatt-hour of electricity generated:
| Energy Source | Deaths per TWh |
|---|---|
| Coal | 24.6 |
| Oil | 18.4 |
| Natural gas | 2.8 |
| Biomass | 4.6 |
| Rooftop solar | 0.44 |
| Wind | 0.15 |
| Hydropower | 1.4 |
| Nuclear | 0.07 |
Hannah Ritchie and Max Roser at Our World in Data, using data from peer-reviewed sources and the Global Burden of Disease study, produce similar estimates. Their analysis, citing Markandya and Wilkinson (2007) in The Lancet, puts nuclear at approximately 0.07 deaths/TWh and coal at approximately 24.6 deaths/TWh — a factor of roughly 350.
The difference between nuclear and coal is not a rounding error. It represents the consequence of choosing, at scale, coal over nuclear. For every terawatt-hour of electricity generated by coal instead of nuclear, approximately 24 additional people die.
The catastrophic accidents are appropriately included in these calculations. Chernobyl, the most severe nuclear accident in history, presents the most complex casualty accounting:
Even using the higher end of estimates (30,000–60,000), Chernobyl's total death toll over 70 years of nuclear power generation globally is comparable to what coal causes globally in a few weeks.
Fukushima caused zero confirmed radiation-related deaths among the public (UNSCEAR 2020 report). The approximately 2,000 deaths associated with Fukushima were caused by the evacuation itself — the stress, displacement, and disruption of vulnerable populations moved from nursing homes and hospitals in the wake of the accident.
The comparison becomes more stark when nuclear's counterfactual is fossil fuels. Air pollution from fossil fuel combustion — not accidents — is the dominant cause of energy-related mortality. The Global Burden of Disease study estimated approximately 8.7 million premature deaths annually from outdoor air pollution, the majority from fossil fuel combustion.
Lelieveld et al. (2019) in European Heart Journal estimated that 790,000 annual deaths in Europe are attributable to fossil fuel air pollution, with the majority from coal combustion. These deaths are diffuse, statistical, and do not generate news coverage. They do not trigger the same response as a nuclear accident that directly kills far fewer people.
Germany's post-Fukushima nuclear phase-out provides a natural experiment. Jarvis, Deschenes, and Jha (2022), in a Nature Energy paper, estimated that the nuclear shutdowns in Germany resulted in approximately 1,100 additional deaths per year from increased fossil fuel combustion and associated air pollution. The total cost over the phase-out period, by their estimates, was approximately $12 billion annually in health costs, primarily from air pollution mortality.
The phase-out did not make Germany safer. It made Germany substantially less safe, by replacing a low-mortality energy source with higher-mortality ones.
Critics of nuclear energy raise several concerns that are not fully addressed by the mortality statistics:
Long-lived radioactive waste. The spent fuel produced by nuclear plants remains radioactive for tens of thousands of years, and no country has yet opened a permanent deep geological repository for high-level nuclear waste. This is a genuine unresolved problem, though the waste volumes are smaller than most people realize (all US commercial nuclear waste ever produced would fit in an area the size of a football field to a depth of about 10 yards).
Proliferation risk. Nuclear power technology can be adapted for weapons purposes. The spread of enrichment and reprocessing capacity creates non-proliferation risks that mortality statistics do not capture.
Tail risk. Some analysts argue that nuclear's very low average mortality masks a high-variance distribution — that the catastrophic tail risk (a worst-case Chernobyl or worse) is systematically underestimated. This is harder to evaluate statistically given the small number of major accidents.
These concerns are legitimate. The point of the mortality comparison is not that nuclear is without downsides, but that the magnitude of nuclear's safety advantage over fossil fuels is large enough that these downsides need to be weighed against it honestly.
By the most directly relevant measure — deaths per unit of electricity generated — nuclear power is among the safest energy sources available, comparable to wind and solar and roughly 350 times safer than coal. The catastrophic accidents that dominate public perception (Chernobyl, Fukushima) are appropriately included in these calculations and do not alter the conclusion.
Public risk perception of nuclear power is inverted relative to the evidence. This inversion has measurable costs: countries that phased out nuclear in response to public fear shifted to higher-mortality alternatives. If energy policy aims to minimize harm, the evidence is clear about the relative safety of nuclear power — even setting aside its zero-carbon benefits. The policy question is how to weigh confirmed mortality advantages against uncertain long-run risks from waste and proliferation. That is a legitimate debate. The claim that nuclear is unusually dangerous is not.
nonacademicresearch.org Editorial (2026). Nuclear Power's Safety Record: What the Evidence Shows. nonacademicresearch.org. nar:8ewi0u3wspa9ha5bur
@misc{e1z9fcjz,
title = {Nuclear Power's Safety Record: What the Evidence Shows},
author = {nonacademicresearch.org Editorial},
year = {2026},
howpublished = {nonacademicresearch.org},
note = {nar:8ewi0u3wspa9ha5bur},
}Temporary identifier. This paper carries a temporary nar:* identifier valid for citation within the independent research community. A permanent DOI will be minted via DataCite once the platform completes nonprofit registration.
Log in to join the discussion.
Loading…