The US Environmental Protection Agency attributes approximately 21,000 lung-cancer deaths per year in the United States to indoor radon exposure, making radon the second leading cause of lung cancer after cigarette smoking and the leading environmental cause (EPA-402-R-03-003, Assessment of Risks from Radon in Homes). About 2,900 of those deaths occur in people who have never smoked; the remainder reflect the strongly multiplicative interaction between radon exposure and tobacco smoke, quantified in the EPA risk tables.
Where the 21,000-deaths figure actually comes from
The headline number is not an epidemiological count. It is a model-based estimate published in 2003 by the EPA in Assessment of Risks from Radon in Homes (EPA-402-R-03-003). The estimate has three layers stacked on top of each other, and understanding the stack is the difference between citing the figure and understanding it.
The first layer is the dose-response model: how much excess lung cancer is expected per unit of cumulative radon exposure. EPA-402-R-03-003 uses the BEIR VI risk model published by the National Research Council in 1999 (Health Effects of Exposure to Radon: BEIR VI). BEIR VI was the consensus US-government dose-response model at the time the EPA risk assessment was written and remains the model that EPA cites today.
The second layer is the exposure distribution: how much radon US homes actually contain. The EPA combined the National Residential Radon Survey (EPA 402-R-92-011), the State/EPA Residential Radon Survey, and supporting state-level data to estimate that the average indoor radon concentration across US homes is roughly 1.25 pCi/L, with a long right tail of homes well above the 4 pCi/L action level. See our explainer on the EPA radon zones map for how this geographic distribution is summarised at the county level.
The third layer is the population: how many people live in those homes, with what smoking histories, for how many years. EPA-402-R-03-003 applies US Census housing and population data and CDC smoking-prevalence data to translate the dose-response and exposure distribution into an expected number of lung-cancer deaths per year.
Run that calculation, and the central estimate is approximately 21,000 lung-cancer deaths per year attributable to radon in US residences. The same assessment estimates roughly 2,900 of those deaths occur in never-smokers; the rest occur in current or former smokers, where radon and tobacco exposures combine multiplicatively (see the smoking-interaction section below).
The miner cohort foundation: BEIR VI
Modern residential radon risk modelling is built on the epidemiology of underground hard-rock miners, where occupational radon exposures were high enough to produce statistically clean dose-response signals. BEIR VI pooled 11 cohorts of underground miners — uranium miners in the United States (Colorado Plateau), Canada, Czechoslovakia, France, and Australia, plus tin and iron miners elsewhere — for a combined cohort of roughly 68,000 miners and approximately 2,700 lung-cancer deaths.
From that pooled data, BEIR VI derived an excess relative risk per working level month (WLM), the standard occupational unit for cumulative radon-progeny exposure. The committee then extrapolated downward to residential exposures, which are typically two to three orders of magnitude lower than the historic mining exposures. Two key features of the BEIR VI extrapolation are worth understanding because they shape every downstream EPA number:
- The model is linear with no threshold over the residential exposure range. Risk per unit of exposure is assumed constant; halving the exposure halves the excess risk.
- The model is sub-multiplicative with smoking. The combined risk for a smoker exposed to radon is larger than the sum of the separate risks but smaller than their pure product — closer to multiplicative than additive, but not exactly either.
The downward extrapolation from mining to residential doses has always been the most-debated step in the analysis, which is why the residential case-control epidemiology that followed BEIR VI matters.
The residential validation: Iowa, Europe, North America
Three sets of residential studies have tested the BEIR VI extrapolation directly, using lung-cancer cases and controls drawn from ordinary households rather than mines.
The Iowa Radon Lung Cancer Study
Field and colleagues published the Iowa Radon Lung Cancer Study in the American Journal of Epidemiology in 2000 (volume 151, issue 11, pages 1091–1102). The study enrolled 413 lung-cancer cases and 614 controls, all women aged 40–84 who had lived in their current Iowa home for at least 20 years. Iowa was chosen because it has the highest mean indoor radon levels of any US state, giving the study statistical power that lower-exposure populations cannot achieve.
The headline result: a roughly 50% increase in lung-cancer odds per 100 Bq/m³ (approximately 2.7 pCi/L) of cumulative residential radon exposure, after adjustment for age, smoking, and occupational exposures. The slope is statistically consistent with the BEIR VI extrapolation from miners.
The European pooled analysis (Darby et al., 2005)
Darby and colleagues published a pooled analysis of 13 European residential case-control studies in the BMJ in 2005 (volume 330, issue 7485, page 223). The combined dataset is 7,148 lung-cancer cases and 14,208 controls from nine European countries, by far the largest residential radon dataset assembled. The pooled estimate is a 16% increase in lung-cancer risk per 100 Bq/m³ of measured radon, rising to roughly 30% after correction for measurement error in the residential exposure assessment. The dose-response was linear with no apparent threshold across the range studied, including concentrations well below the EPA action level.
The North American pooled analysis (Krewski et al., 2005)
Krewski and colleagues published a pooled analysis of seven North American residential case-control studies in Epidemiology in 2005 (volume 16, issue 2, pages 137–145), drawing on US and Canadian datasets including Iowa, Missouri, New Jersey, Connecticut, and Winnipeg. The pooled excess odds ratio was approximately 11% per 100 Bq/m³, again with no evidence of a threshold and broadly consistent with both the European and miner-derived estimates.
The three lines of evidence — miner cohorts (BEIR VI), the Iowa study, and the European and North American pooled analyses — converge on the same conclusion: residential radon causes lung cancer, the dose-response is linear without a clear threshold, and the BEIR VI slope used by EPA-402-R-03-003 is a defensible model for residential exposures.
The smoking interaction: why the multiplier matters
BEIR VI and EPA-402-R-03-003 model the joint effect of radon and tobacco smoke as roughly multiplicative. The practical consequence is that radon's absolute risk depends heavily on whether the person exposed has ever smoked. EPA's published Table A-1 quantifies this directly: at every exposure level, the risk to a current or former smoker is approximately an order of magnitude higher than the risk to a never-smoker at the same concentration.
That multiplier is the single most important policy implication of the EPA risk assessment. The same physical radon level in the same room produces dramatically different lifetime-mortality estimates depending on the occupant's smoking status, and the largest absolute reduction in radon-attributable deaths therefore comes from mitigating the homes of current and former smokers.
EPA Table A-1: lifetime lung-cancer risk by concentration and smoking status
The table below summarises the EPA-published lifetime risk estimates from Table A-1 of EPA-402-R-03-003, expressed as the number of additional lung-cancer deaths per 1,000 people exposed for a lifetime (taken to be 70 years) at the listed indoor radon concentration. Numbers are rounded to two significant figures for readability; consult the EPA document for unrounded values and confidence intervals.
| Indoor radon (pCi/L) | Never-smokers (per 1,000) | Current/former smokers (per 1,000) | Context |
|---|---|---|---|
| 20 | ~36 | ~260 | 5× the EPA action level |
| 10 | ~18 | ~150 | 2.5× the action level |
| 8 | ~15 | ~120 | 2× the action level |
| 4 | ~7 | ~62 | EPA action level |
| 2 | ~4 | ~32 | EPA "consider action" threshold |
| 1.25 | ~2 | ~20 | US indoor average |
| 0.4 | <1 | ~7 | US outdoor average |
Read the columns horizontally and the smoking multiplier is obvious: at 4 pCi/L, the never-smoker estimate is roughly 7 lifetime deaths per 1,000 people, while the smoker estimate is roughly 62 per 1,000 — a multiplier of about 9. The pattern holds across the full table.
Read the columns vertically and the dose-response is equally clear: doubling the radon concentration roughly doubles the modelled lifetime risk in both columns. Cutting the concentration in half — the difference between, say, mitigating from 8 pCi/L down to 4 pCi/L — roughly halves the radon-attributable lifetime risk for everyone in the home.
How to read these numbers responsibly
These are population lifetime risks under a stylised exposure scenario — 70 years of continuous exposure at the listed concentration. They are not personal predictions, and they cannot be translated into an individual probability that any particular person will get lung cancer. Use them to compare exposure levels and inform mitigation decisions, not to forecast your own outcome. For health-status questions, consult a physician.
What the table is — and what it is not
EPA Table A-1 is a policy-grade comparative tool. It exists so that homeowners, real-estate professionals, mitigators, and state radon programmes can talk about radon risk in a common, EPA-sanctioned vocabulary. It has three properties worth keeping in mind:
- It is a population estimate, not an individual prediction. The risk attributable to radon depends on the joint distribution of exposure, smoking, age, gender, and competing causes of death; the table averages across that joint distribution.
- It assumes lifetime exposure. The 70-year scenario is unrealistic for most homeowners, who move multiple times and whose homes' radon concentrations vary by season and by changes in the building envelope. Real-world cumulative exposure is almost always a weighted average over several homes.
- It is anchored to the BEIR VI dose-response. If a future risk model revises BEIR VI — either up or down — the table will move accordingly.
None of these caveats undermines the core finding. They simply explain why the EPA presents the numbers as comparative risk estimates rather than personal probabilities.
Is there a threshold below which radon is safe?
The short answer in current US and international guidance: no demonstrated threshold has been observed. BEIR VI, the European pooled analysis (Darby et al., 2005), and the North American pooled analysis (Krewski et al., 2005) all report a linear dose-response with no observable point below which excess risk falls to zero. Risk is lower at lower concentrations, but it is not zero.
This is the reason the EPA action level of 4 pCi/L is described not as a "safe" level but as the concentration at which mitigation is unambiguously recommended. EPA further suggests that homeowners consider mitigation between 2 and 4 pCi/L, recognising that the risk at 2 pCi/L is non-trivial — particularly for smokers. The World Health Organization in its Handbook on Indoor Radon (2009) uses a still lower reference level of 100 Bq/m³ (approximately 2.7 pCi/L), explicitly on the basis that no threshold has been demonstrated.
For the geological and geographic context behind these numbers — why some counties consistently produce higher indoor radon than others — see our companion explainer on the EPA radon zones and the directory of the highest-radon US counties. For homes served by private well water, radon dissolved in groundwater is a secondary indoor source covered in our guide on radon in well water.
How radon compares with other mortality risks
The Surgeon General's 2005 Health Advisory on Radon framed the figure in plain comparative terms: at roughly 21,000 lung-cancer deaths per year, radon-attributable mortality in the United States is on the same order of magnitude as the annual death toll from drunk-driving crashes and exceeds the annual death toll from residential fires, by the figures the Advisory cited from CDC and NHTSA data of that era. The Advisory used the comparison to underline that radon is a preventable environmental hazard with a low-cost mitigation pathway, not to argue any particular numeric equivalence with another risk.
The National Cancer Institute's Radon and Cancer fact sheet states the same conclusion in different words: radon is the leading cause of lung cancer among people who have never smoked and the second leading cause overall, behind cigarette smoking. Both characterisations descend from the same EPA-402-R-03-003 / BEIR VI lineage.
What this means practically for a homeowner
The risk numbers translate cleanly into three operational decisions:
- Test the home. The EPA recommends every home be tested for radon, regardless of geography or construction type. Our DIY radon testing guide walks through which kit to buy and how to deploy it correctly; the trade-offs between short-term charcoal tests and long-term alpha-track detectors are covered in short-term vs. long-term radon testing.
- Mitigate at or above 4 pCi/L, and consider mitigation between 2 and 4 pCi/L. The standard EPA-recommended mitigation technique is sub-slab depressurization, which typically reduces indoor radon by 50–99%. The engineering details are in sub-slab depressurization explained.
- Account for smoking status when ranking priorities. A smoker living in a 6 pCi/L home faces a much larger absolute radon-attributable risk than a never-smoker in the same home (compare the two columns of Table A-1 above). For households with current or former smokers, the relative value of mitigation is correspondingly higher.
From the US Surgeon General's 2005 Health Advisory on Radon
"Indoor radon is the second-leading cause of lung cancer in the United States and breathing it over prolonged periods can present a significant health risk to families all over the country." The Advisory frames radon as a preventable hazard with no known safe level of exposure, and urges every US home below the third floor to be tested.
Anchoring the report to these sources
RadonZoneReport's county-level dossier presents the EPA radon-zone classification and state-survey averages for the county you specify, alongside the EPA risk-table framing above for the concentration range observed. The numbers we cite are from the primary EPA, BEIR VI, and pooled-study publications listed below — we do not generate epidemiological estimates ourselves. The full source list and the publication numbers are on our sources page; the build process and assumptions are documented in the methodology. To pull the dossier for your county now, visit the order page.
Sources
- US Environmental Protection Agency. EPA Assessment of Risks from Radon in Homes. EPA-402-R-03-003, June 2003. The primary federal risk assessment, including Table A-1 of lifetime lung-cancer deaths per 1,000 by exposure level and smoking status.
- National Research Council. Health Effects of Exposure to Radon: BEIR VI. National Academies Press, 1999. The pooled miner-cohort dose-response model on which EPA-402-R-03-003 is built.
- Field, R. W., et al. "Residential Radon Gas Exposure and Lung Cancer: The Iowa Radon Lung Cancer Study." American Journal of Epidemiology 2000; 151(11): 1091–1102.
- Darby, S., et al. "Radon in homes and risk of lung cancer: collaborative analysis of individual data from 13 European case-control studies." BMJ 2005; 330(7485): 223.
- Krewski, D., et al. "Residential radon and risk of lung cancer: a combined analysis of 7 North American case-control studies." Epidemiology 2005; 16(2): 137–145.
- US Surgeon General. Health Advisory: National Health Advisory on Radon. Office of the Surgeon General, January 2005.
- National Cancer Institute. Radon and Cancer fact sheet, cancer.gov.
- World Health Organization. WHO Handbook on Indoor Radon: A Public Health Perspective. WHO, Geneva, 2009.
Related guides
Get the county-specific dossier
The 6-page PDF combines the EPA radon-zone designation, state-survey averages, and the EPA Table A-1 risk framing into one printable page for the county you specify.
Pull my report · $15