Do CT Scans Cause Cancer? How to Weigh the Linear No Threshold Model

CT scans do deliver ionizing radiation, and radiation can cause cancer, so the honest answer is a qualified yes: a single medically justified scan carries a very small risk that is usually outweighed by what the images reveal. The widely reported 2025 projection that CT imaging could drive roughly 103,000 future cancers in the United States is a model output, not a count of real tumors. It comes from feeding 93 million scans into a risk calculator built on the linear no threshold assumption, which presumes that every dose down to zero carries some proportional risk. Knowing where that number is solid and where it is contested is the difference between informed caution and avoidable fear.

Where the 103,000 figure comes from

The projection traces to a study by Rebecca Smith-Bindman and colleagues published in JAMA Internal Medicine in April 2025 and summarized by NIH Research Matters. The team estimated that the roughly 93 million CT examinations performed on about 61.5 million patients in 2023 could eventually produce around 103,000 radiation-induced cancers, with a 90% uncertainty range of about 96,400 to 109,500. If imaging patterns hold, they argued, CT-associated cancers could account for close to 5% of new cancer diagnoses each year. The most commonly projected types were lung, colon, and leukemia. Per-scan risk was higher in children, but because adults are scanned far more often, adults accounted for most of the projected total.

The engine behind these numbers is the National Cancer Institute's Radiation Risk Assessment Tool, or RadRAT, which draws on risk models from the National Academies' BEIR VII report. That lineage matters, because it tells you what kind of number 103,000 is. It is an extrapolation, an estimate of what should happen if a particular dose-risk relationship holds all the way down to the small doses of a single CT.

What the linear no threshold model actually assumes

The linear no threshold model, usually shortened to LNT, holds that cancer risk rises in direct proportion to radiation dose and that there is no threshold below which risk becomes zero. Its strongest evidence comes from populations exposed to relatively high doses, most famously the Japanese atomic-bomb survivors, where excess cancers are clearly measurable. LNT takes that measured relationship and draws a straight line down through the low-dose region where direct measurement is hard.

Radiation-protection bodies such as the International Commission on Radiological Protection adopt LNT as a prudent default for setting exposure limits. Using a conservative, no-safe-dose assumption makes sense when the goal is to keep occupational and public exposure as low as reasonably achievable. The controversy is not whether LNT is a reasonable planning tool. It is whether the same assumption should be used to convert millions of tiny individual exposures into a specific national cancer count.

Modeled estimate versus measured harm

That question sits at the heart of a 2025 viewpoint in the American Journal of Respiratory and Critical Care Medicine, whose title argues plainly that the 5% figure is overstated. The authors contend that the projection is largely a modeling artifact, a theoretical upper bound produced by pushing older radiation-risk models far below the exposure range where their evidence is solid. Their core methodological objection concerns collective dose, the practice of adding up very small per-person risks across a huge population to generate a large aggregate number. ICRP itself cautions against using collective dose this way to compute a body count, precisely because the individual risks being summed sit below the level at which epidemiology can confirm them. Below roughly 100 millisievert, the excess cancer signal is faint and easily swamped by ordinary cancer rates and statistical noise.

None of this means CT radiation is harmless. It means a projection carries the uncertainty of its assumptions, and a figure reported to three significant digits can imply more precision than the underlying science supports.

What the empirical data do show

The strongest measured evidence sits between the two camps. The EPI-CT study, coordinated by the International Agency for Research on Cancer, followed roughly 948,000 people who were scanned before age 22 across nine European countries. Published in 2023, it found a genuine dose-response: the risk of blood cancers rose with cumulative bone-marrow dose, with an excess relative risk near 1.96 per 100 milligray, close to a tripling of risk at that exposure. That is measured harm, not a projection, and it argues against the idea of a clean safe threshold, at least in children, whose dividing tissues are more radiosensitive.

Read carefully, though, EPI-CT describes cumulative dose in a young, sensitive population followed over years. It supports the direction of the LNT assumption without validating any single national headcount for adults, most of whom receive one scan at a time for a specific clinical question.

How to weigh it for yourself

The useful distinction is between the individual and the population. On the day a scan is ordered, the relevant question is whether the images will change what happens next. A CT that confirms or rules out a stroke, a clot, or a tumor delivers a benefit that swamps a per-scan cancer risk usually estimated at a small fraction of a percent. The population number becomes worrying mainly when scans are duplicated, ordered reflexively, or performed when a lower-dose test would answer the same question. Two levers actually reduce risk: justification, meaning the scan should be clinically warranted, and optimization, meaning modern equipment and protocols keep dose as low as the diagnostic task allows.

This is educational information, not medical advice, and decisions about any specific scan belong with you and the clinician who knows your situation. Weighed honestly, that 103,000 figure is most useful as a case against unnecessary imaging. It should sharpen judgment about which scans to order while leaving room to accept a scan you genuinely need.