Living Near A Nuclear Plant Raises Cancer Death Risk, US Study Finds

The map looks almost geological. Across the eastern United States, counties glow dark purple wherever nuclear plants cluster: along the Great Lakes corridor, the mid-Atlantic, the Connecticut River valley. Head west past the Mississippi and the shading bleaches out to pale yellow. Fewer plants, fewer people caught in their gravitational pull.

That map is also, according to a new Harvard study published today in Nature Communications, a rough approximation of elevated cancer risk.

Researchers at Harvard T.H. Chan School of Public Health spent years assembling what is, by their reckoning, the most comprehensive national picture yet of cancer mortality and proximity to nuclear power plants across the United States. Their conclusion is uncomfortable, particularly given the moment we’re in: counties closer to operational nuclear plants have higher cancer death rates than those further away—and the association holds even after you strip out poverty, smoking, obesity, race, healthcare access, and most other plausible explanations.

The numbers the team arrived at are striking. Over their 19-year study period (2000 to 2018), they attribute roughly 115,000 cancer deaths—about 6,400 a year—to proximity to nuclear facilities. That figure comes with wide confidence intervals, and the researchers are emphatic that correlation is not causation. But it’s not nothing. For context, a major 2023 Science paper estimated that coal power plant emissions cause around 20,900 all-cause deaths annually. The nuclear cancer figure, uncertain as it is, works out to roughly 20% of that coal mortality burden—and this counts cancer deaths only.

The team, led by doctoral researcher Yazan Alwadi and senior author Petros Koutrakis, took a different methodological approach from most previous work in this area. Earlier U.S. studies tended to focus on a single plant and its surrounding community, which limits statistical power and forces binary choices: you’re either inside some arbitrary radius or you’re not. Alwadi and colleagues instead used what they call continuous proximity—a running tally of inverse distances to every operational nuclear plant within 200 km of each county’s centre, summed across all plants and averaged over 10-year windows to account for the long latency periods typical of radiation-related cancers.

That matters. A county wedged between two plants might score high on this metric even if it’s not especially close to either one. The approach is more realistic about how cumulative environmental exposures actually work.

The results were consistent across multiple sensitivity checks. When the team varied the distance cutoff—from 200 km down to 100 km, in 10 km steps—the associations held. When they tried different averaging windows (2, 5, 10, 15, 20 years), the findings remained stable. These aren’t the fragile results that evaporate when you poke at the assumptions.

Older adults are the group most at risk, and the age-sex patterns are interesting. The peak associations appear a decade apart by sex: the highest relative cancer mortality risk for females sits in the 55–64 age group, while for males it lands 65–74. This kind of sex-differential timing is consistent with known differences in radiation sensitivity and cancer biology, though the study wasn’t designed to disentangle that. At the closest equivalent distances, the relative risk for the highest-exposure groups reaches about 1.20—a 20% elevation compared to the most distant counties.

“Our study suggests that living near a NPP may carry a measurable cancer risk—one that lessens with distance,” says Koutrakis, the Akira Yamaguchi Professor of Environmental Health and Human Habitation at Harvard. “We recommend that more studies be done that address the issue of NPPs and health impacts, particularly at a time when nuclear power is being promoted as a clean solution to climate change.”

That last clause is where the finding becomes politically charged. Nuclear energy is experiencing something of a renaissance in climate discussions—new builds, reactor restarts, government subsidies—precisely because it generates electricity without carbon emissions. The idea that proximity to plants might carry a cancer cost sits awkwardly alongside that narrative.

To be clear about what this study can and cannot show: it doesn’t measure actual radiation doses. It assumes, perhaps unrealistically, that all plants contribute equally to exposure regardless of design, age, or operating practices. The county-level data can’t track where individuals actually lived or moved. And because it’s ecological—populations, not people—it can’t rule out unmeasured confounders doing the work.

The research team is candid about all of this. The study explicitly notes that its exposure metric reflects geographic proximity rather than actual radiation experienced by individuals, and that it cannot speak to specific cancer types, mechanisms, or the childhood cancer question, which requires different methods entirely.

The global literature on nuclear plants and cancer is genuinely messy. Some national studies—in the US, UK, and Canada—have found nothing. Others, in France, Spain, and South Korea, have found elevated risks for specific cancers near specific plants. A German study found leukemia rates more than twice as high in children under five living within 5 km of a plant. The inconsistency isn’t necessarily suspicious; it likely reflects real differences in plant design, emissions controls, and crucially, in how studies are designed and what they can statistically detect.

What this new work adds is scale and methodological refinement. Single-plant studies are underpowered. National studies with fixed-radius cutoffs oversimplify exposure. By modelling continuous proximity across all plants and all counties over nearly two decades, this team had more statistical muscle to work with—and found something.

Ionizing radiation is an established carcinogen. We’ve known that since Hiroshima, refined it through decades of studying atomic bomb survivors, and confirmed it from Chernobyl, however contested those post-accident findings remain. That low-level chronic exposure near operating civilian plants could contribute to cancer is biologically plausible. The question has always been whether the doses involved are large enough to produce detectable signal at the population level.

This study suggests the answer might be yes. The public health arithmetic, even at the lower bounds of the confidence intervals, is not trivial. Roughly 4,266 deaths per year among people aged 65 and above, associated with NPP proximity—that’s the kind of number that ordinarily prompts regulatory interest.

What happens next is uncertain. The researchers want more investigation: finer-grained exposure data, individual-level cohort studies, dose reconstruction efforts that could actually tie cancer outcomes to measured radiation rather than map coordinates. These are harder studies to run but more defensible ones.

For now, millions of Americans live within the purple zones on that map. Most of them are also, probably, fine—any individual excess risk remains modest, even at the closest distances. But “probably fine” and “no measurable population effect” are different things, and this study is arguing they’re not the same.

As nuclear energy positions itself as a pillar of decarbonisation, questions about what it does to the people who live next door deserve better answers than we currently have.

Study link: https://www.nature.com/articles/s41467-026-69285-4