Research Article

Nuclear Accidents: real consequences and persistent myths

Author
Juan Beliera
Published
May 29, 2025

What should we expect from a nuclear accident?

The worst-case scenario in a nuclear accident is a partial or complete meltdown of the reactor core. In such events, it's highly likely that significant amounts of radiation are released into the environment. This radiation exposure can lead to two main types of consequences:

  • At very high radiation levels, acute radiation syndrome (ARS) may develop, potentially leading to death within days or weeks.

  • At lower levels, there's an increased probability of developing cancer over the course of years or even decades.

In most nuclear accidents, the primary health concern falls into the second category: a portion of the population receives doses of radiation that may increase cancer risk over the medium to long term.

As discussed in a previous article, the likelihood of such an event in modern nuclear facilities has been reduced so dramatically that it's now statistically more probable to die from an accident related to solar or wind energy production than from one involving a nuclear reactor.

Moreover, as we’ll elaborate in a future article, modern plants are designed with additional safety layers to ensure that, even in the rare event of a severe accident, the off-site consequences remain significantly lower than what would be expected from older reactor designs.

What major nuclear accidents have occurred?

To date, there have been three major nuclear accidents that resulted in off-site radiation releases. The directly attributable health impacts are summarized as follows:

  • Three Mile Island (EEUU, 1979)
    • No fatalities.
  • Chernobyl (URSS, 1986)
    • 2 deaths from a steam explosion.
    • 28 fatalities from acute radiation syndrome (ARS).
    • ~5,000 cases of thyroid cancer.
    • No detectable increase in overall cancer incidence in Ukraine, Russia, or Belarus.
    • WHO estimates up to 4,000 premature deaths potentially attributable to radiation exposure.
  • Fukushima (Japan, 2011)
    • No radiation-related fatalities

These figures are based on decades of rigorous epidemiological studies involving populations exposed to the fallout from each event. They should not be interpreted as minimizing the significance of nuclear accidents but rather as providing an accurate perspective—the actual death tolls are substantially lower than what is often assumed in the public imagination.

It's also worth highlighting that none of the accidents in reactors built outside the former Soviet Union resulted in radiation-related fatalities. This reflects the markedly lower safety standards in the Soviet bloc at the time—standards that have since been brought closer to international norms.

Finally, the psychological and social consequences, such as the trauma of large-scale evacuations and the effects of misinformation, must also be taken into account when evaluating the full impact of these events.

But what about estimates of hundreds of thousands of deaths from Chernobyl?

Such claims are often based on flawed methodological approaches. The most common issue lies in the misuse of the Linear No-Threshold (LNT) model, a conservative framework used for radiation protection purposes.

What is the linear no-threshold model?

To regulate acceptable levels of radiation exposure in the general population, a conservative approach known as the Linear No-Threshold model (LNT) is commonly used.

This approach stems from the inherent limitations in accurately measuring the natural incidence rate of cancer and how it changes in populations exposed to radiation. There are several reasons for this:

  • The variations in cancer rates due to low radiation exposure are small compared to the baseline incidence rate.

  • The natural incidence of cancer has significant intrinsic variability.

  • In general, cancer incidence rates are poorly measured—particularly in countries with less developed healthcare systems, and especially when examining historical data from earlier decades.

As a result, the most reliable data linking radiation exposure and cancer incidence come from events where radiation levels were extremely high. In such cases, statistical variations are more easily detected. The most commonly cited example is the long-term health consequences of the atomic bombings in Japan during World War II.

Taking these limitations into account, the LNT model is based on two main assumptions:

  1. The biological effect of receiving a high dose of radiation in a short period is equivalent to receiving the same dose spread out over a longer period.

  2. There is no threshold dose of radiation below which there is zero risk to human health.

Both assumptions are intentionally conservative, and there is growing evidence suggesting that they may not be entirely accurate. However, to date, the LNT model remains the best available framework for establishing radiation protection standards in the nuclear sector.

Misapplication of the model

The common misuse goes like this:

  • Estimate the total radiation released during a nuclear accident (so far, so good).
  • Estimate the population exposed to that radiation (still valid).
  • Apply the LNT model to calculate an increase in cancer incidence (this is where the problem arises).

This method produces high fatality estimates because it uses a model designed for regulatory conservatism, not for post-accident epidemiological assessment.

Let’s use an analogy:

  • A bridge is designed to support 1,000 kg.
  • To account for uncertainty, engineers include a safety factor of 1.5.
  • A 1,100 kg vehicle crosses the bridge.
  • If one applies the design assumptions rigidly, they might conclude the bridge has failed.
  • Yet, upon inspection, the bridge remains intact—clearly, the conservative model doesn’t reflect the real-world outcome.

In both cases—the nuclear accident and the bridge—the correct way to assess the consequences of an event is to measure what actually occurs.

Giving equal or greater weight to conservative design models than to empirical findings from post-accident epidemiological studies is a methodological error.

How does nuclear power compare to other industries?

Table 1 presents a list of major industrial disasters from various sectors. When we compare these events to nuclear accidents, it's clear why the fatality rate per unit of energy—as shown in a previous article—is much lower for nuclear power.

Event Year Country Fatalities
Banqiao Dam Failure 1975 China 100000–250000
Machchhu Dam Failure 1979 India 5000–10000
The great smog of London 1952 UK 4000–6000
Train derailment (Tsunami) 2004 Sri Lanka 1700
Bhopal pesticide plant disaster 1984 India 4000–16000
San Juanico disaster 1984 Mexico 500
Piper Alpha disaster 1988 Scotland 167
Guadalajara explosions 1992 Mexico 206
Mumbai High oil fire 2005 India 22
Lac-Mégantic oil train explosion 2013 Canadá 47
Brumadinho dam disaster 2019 Brazil 259

Final Thought

When considered in the broader context of industrial risk, nuclear energy is not only one of the cleanest energy sources but also among the safest—even when accounting for the worst accidents in history.

Understanding the true consequences of nuclear accidents—and distinguishing them from enduring myths—is essential if we're to engage in informed discussions about the future of energy.

Stay tuned for our next article, where we’ll take a closer look at each of the three major nuclear accidents: their technical causes, health impacts, and the public perceptions they shaped.