Bananas, Airplanes, and CT Scans: Putting Radiation in Perspective

Most people worry about the wrong exposures, and not enough about the ones that quietly add up.

Patients ask me about radiation more often than almost any other safety question. Is a mammogram going to give me cancer? Should I refuse the follow-up CT? Is it safe to fly? The honest answer is that the fear is usually misdirected. The exposures that frighten people most are almost always trivial, and the ones that genuinely deserve attention rarely get counted. What follows is an attempt to give you the actual numbers, in plain language, along with an honest account of where the science is settled and where it is not.

First, what a millisievert is

Radiation dose gets measured in millisieverts, abbreviated mSv. A sievert captures not just how much radiation energy your tissues absorb, but how much biological harm that particular kind of radiation is expected to do. A whole sievert is enormous, so everyday exposures are quoted in thousandths of one.

A useful trick for making these numbers intuitive is to convert them into flight-hours. Up at 10,000 meters you sit above much of the atmosphere that normally shields you from cosmic rays, so you pick up roughly 0.004 mSv for every hour in the air. Once you have that conversion, the scale of things becomes obvious. An hour of flying, a panoramic dental film, and sleeping next to another (very slightly radioactive) human being are all roughly interchangeable. You already accept small doses constantly and never think about them.

One more term is worth knowing. Effective dose is a whole-body average that lets us compare a chest x-ray with a bone scan even though they irradiate completely different tissues. It is a planning and comparison tool, not a measurement of what happened inside your particular body. That distinction will matter later.

The everyday exposures are genuinely trivial

An airport body scan, an hour of sunshine (the ionizing component, not the ultraviolet that actually causes skin cancer), a household smoke detector, living near a nuclear generating station: all of these land somewhere in the range of thousandths or ten-thousandths of a millisievert. These are doses your cells were designed to handle, repairing occasional molecular damage continuously as ordinary housekeeping. We are not fragile in this respect. We are remarkably well engineered.

Two of these deserve comment. The banana is the most famous small dose in the world, and the figure usually quoted, about 0.0001 mSv, is right. But radiologists dislike the so-called banana equivalent dose for a good reason. The radioactivity comes from potassium-40, and your body holds potassium at a tightly regulated level. Eat a banana, and your kidneys quietly excrete the surplus within hours. The banana illustrates how small a fraction of a millisievert is, but it is a poor model for doses that actually deposit and stay.

The second is a genuine surprise to most people: living near a coal-generating station delivers more radiation dose than living near a nuclear one. Coal contains traces of uranium and thorium, and burning it concentrates them in fly ash. It is a good reminder that intuitions about radiation and intuitions about actual risk often point in opposite directions.

For scale, natural background radiation runs about 2-3 mSv per year worldwide, varying considerably by geology and altitude, and the average American receives roughly 6 mSv annually once medical imaging is included. Everything in the previous two paragraphs is a rounding error against that baseline.

The exposure where averages deceive

Smoking is sometimes listed at around 0.18 mSv per year for half a pack a day, which makes it look like another rounding error. That figure is a whole-body effective dose, and here the whole-body average hides the real story.

The polonium-210 and lead-210 in tobacco smoke do not distribute evenly. They lodge at specific branch points in the airways and irradiate a small patch of bronchial tissue at very close range, year after year. Estimates of the localized dose to those spots run into the tens of millisieverts annually for a moderate smoker, orders of magnitude above the averaged number. This is one reason radiation is thought to contribute to smokers’ lung cancer alongside the chemical carcinogens, and it is a useful lesson in why a single averaged figure can mislead badly.

X-rays and mammograms

A chest X-ray costs you about 0.1 mSv. An arm film or a dental panoramic view is smaller still. These are trivial, amounting to a few days or a few weeks of the natural background you receive simply by existing. A screening mammogram runs roughly 0.4 mSv, which is also modest, and that figure holds up well against current digital mammography.

Nobody should decline any of these tests over radiation worry. I want to be as unambiguous as I can be, because I have watched patients talk themselves out of imaging that would have caught something early, based on a vague sense that X-rays are dangerous.

CT is where the numbers start to matter

Computed tomography is the workhorse of modern diagnosis and also the dose-heavy hitter. Typical averages run around 2 mSv for a head CT, 6 mSv for a chest or pelvic study, and 8 mSv for an abdomen, with the standard published catalog placing most CT examinations somewhere in the 2-20 mSv range (Mettler et al.). Two things about those numbers deserve more attention than they get.

The first is encouraging. Modern dose-reduction methods, especially iterative and now deep-learning image reconstruction, have substantially reduced many CT doses over the past decade. A scan performed today may cost you meaningfully less than the identical scan a decade ago.

The second is not encouraging. The dose for the same examination varies enormously between facilities. In a landmark study of Bay Area institutions, Smith-Bindman and colleagues found roughly a thirteen-fold spread between the highest and lowest dose for each study type, a gap that patient size could not explain. Where you are scanned can matter as much as what is scanned. Most patients have no idea this variation exists, and it is one of the few radiation facts that is genuinely actionable.

The scans nobody counts

Nuclear medicine is the blind spot in most patients’ mental arithmetic, and it matters enormously, because these are exactly the studies my oncology patients receive.

A combined PET/CT scan, routine in cancer staging and surveillance, typically delivers 14-25 mSv in total, with the CT portion usually the larger contributor. A nuclear cardiac stress test, or myocardial perfusion imaging, has historically ranged from 13-16 mSv, although modern cameras and stress-only protocols can bring a well-chosen study down to a few millisieverts. A bone scan adds several more. A patient can easily believe that a CT is the worst they will ever face, when the nuclear study nobody described in dose terms delivered considerably more.

Children are the other category deserving separate treatment. They are more radiosensitive both because their tissues are dividing more actively and because they have more decades ahead in which a cancer could surface. That is precisely why pediatric imaging warrants child-sized protocols and extra scrutiny.

The figure everyone misreads

You will often see 100 mSv described as the lowest acute dose known to cause cancer. It is the single most misunderstood number in this entire subject, and it is worth stating plainly what it does and does not mean.

It does not mean that doses below 100 mSv are safe. It means that 100 mSv is approximately the lowest dose at which epidemiology has enough statistical power to reliably detect an excess of cancers above the normal background rate. Cancer is common. Finding a small extra signal inside a large ordinary number requires enormous study populations, and the required number rises steeply as the dose falls. Below roughly 100 mSv the expected effect is buried in statistical noise, so studies can neither confirm it nor rule it out. Absence of proof is not proof of absence, and that phrase is doing real work here.

What the science actually says about low doses

We are more certain than the reassurers admit and less certain than the alarmists claim.

Above roughly 100 mSv delivered quickly, there is no serious debate. Radiation causes cancer, as the atomic bomb survivor cohorts established decades ago. The real controversy lies entirely in the low-dose range, where medical imaging lives.

Most regulators worldwide use what is called the linear no-threshold model, which assumes risk continues in a straight line down to zero dose with no safe cutoff. It is deliberately cautious, and two large recent studies have strengthened it. The INWORKS study followed more than 300,000 nuclear workers across three countries and found solid cancer mortality rising with cumulative dose, importantly among workers whose exposures were low and spread thinly across many years (Richardson et al.). The multinational EPI-CT study of nearly a million young Europeans found blood cancer risk tracking with CT radiation dose to the bone marrow, translating to roughly 1-2 extra cases per 10,000 children scanned over the following 12 years (Bosch de Basea et al.). A companion analysis found a similar dose-response signal for brain cancer (Hauptmann et al.).

In fairness, a minority of respected scientists disagree. They point to reanalyses of the survivor data suggesting a possible threshold, to laboratory evidence that cells repair low-dose damage efficiently, and to the hypothesis that very small doses might even be mildly protective by activating those repair systems. They argue that applying a no-threshold assumption to trivial doses fuels a radiophobia that harms patients. This is a live scientific debate rather than settled dogma, and you deserve to know that instead of being handed a single confident answer.

Where does this leave you? In a sensible middle. The risk from any single low-dose scan is very small, possibly zero, and certainly far smaller than the benefit of a scan you actually need. But a very small risk multiplied across a hundred million scans is not zero at the population level, which is why minimizing unnecessary dose remains the right policy.

That arithmetic is exactly what a 2025 analysis in JAMA Internal Medicine performed. Projecting from current practice, Smith-Bindman and colleagues estimated that the 93 million CT examinations performed in the United States in 2023 could eventually lead to roughly 103,000 future cancers, and that if practice does not change, CT-associated cancers could account for about 5% of new cancer diagnoses annually. Read that carefully. It is a model built on the no-threshold assumption, not a body count, and several radiologists have publicly criticized it as needlessly frightening. But even taken cautiously, it makes the central point: CT is enormously beneficial, enormously common, and small per-scan risks multiplied by very large numbers deserve attention at the health-system level.

For cancer patients, cumulative dose is the real question

For any individual, the number that matters is not any single scan but the total accumulated across years. This is where my own patients live.

Studies using automated dose tracking have found that a small but real fraction of imaged patients cross 100 mSv of cumulative effective dose from CT alone, usually within just a few years, and usually because they have cancer, serious chronic disease, or major trauma. Surveillance imaging is the common thread. Add PET/CT to a recurring schedule, and the total climbs faster than most patients, and frankly most clinicians, appreciate, because nobody is keeping a running tally.

This is the group for whom lower-dose protocols genuinely matter, and for whom substituting ultrasound or MRI where the clinical question can be answered equally well is worth a real conversation rather than a reflex.

Some benchmarks worth carrying around

Flight crew average around 3 mSv per year from cosmic rays, roughly what an ordinary person receives from natural background alone. Fukushima recovery workers averaged about 12 mSv in the first year, well below any level causing acute harm, though a handful of individuals received far more. An astronaut on a six-month space station mission receives about 72 mSv. The occupational dose limit for radiation workers is 50 mSv in a year. Each of these dwarfs a chest X-ray, and informed adults in specific roles accept them knowingly.

One benchmark needs clarification, and it trips up nearly everyone who encounters it. The public dose limit of 1 mSv per year applies to man-made sources above natural background, such as emissions from a nuclear facility, for members of the general public. It does not apply to the medical radiation a patient receives for their own care. Your CT scan is not over the legal limit. That limit was never written to govern justified medical exposure, and confusing the two creates unnecessary fear.

What to actually do with all this

Before any test that uses radiation, ask three questions. Will the result change my treatment or my doctor’s decision? Is there a test without radiation, such as ultrasound or MRI, that would answer the same question just as well? And can this be done with a low-dose protocol? A good clinician will welcome all three. The principles radiologists themselves work by are justification, meaning the scan should do more good than harm, and optimization, meaning use the lowest dose that still answers the question. The Image Gently campaign for children and Image Wisely for adults exist to keep those principles in front of busy practices.

Keep your own imaging record. Note the date, scan type, and facility for every CT, PET/CT, nuclear medicine study, and fluoroscopy procedure. You do not need to track plain films or dental x-rays, which are too small to matter. This record becomes genuinely useful when you see multiple specialists who cannot see each other’s orders, and it helps you avoid the duplicate scan nobody meant to order.

Match your concern to the dose. Decline nothing over an X-ray, a dental film, or a mammogram on radiation grounds. Save the real discussion for repeated CT, PET/CT, and nuclear cardiac studies, particularly in children and young adults, and particularly when they are scheduled to repeat indefinitely.

And please do not let radiation fear cause the larger harm. The reverse error is real and well documented. Patients who refuse or delay a needed scan, or who accept an inferior test to avoid radiation, sometimes end up with a missed or late diagnosis. That harm is usually far more likely than the hypothetical harm from the dose. A scan your physician can justify is almost always worth its radiation.

References

Bananas, Airplanes, and CT Scans: Putting Radiation in Perspective

Dr. Daniel Thomas, DO, MS