One in a Million, With a Name: How the Micromort Measures Death
A micromort is a one-in-a-million chance of dying, packaged as a unit. Giving tiny risks a shared scale lets them be compared and, within limits, added up. And the limits turn out to be the interesting part.
A metre is a length with a name.
Strictly speaking, nobody needs the name. You could describe every distance as some fraction of the walk from your front door to the shops. But you'd struggle to compare notes with anyone else, and adding up a journey would be a mess.
The micromort does the same job for something rather less pleasant. It's a one-in-a-million chance of death, given a name so that it can be counted.
A slice of a chance
Start with a more general idea, the microprobability: a one-in-a-million chance of anything at all. A micromort is that same slice applied to one particular outcome, dying. Scale it up and the language keeps going. A million micromorts make a mort, which means certain death. So the unit runs from one (very nearly nothing) to a million (very definitely something).
Why bother? Consider a raw count of deaths. In one year's figures, one activity was linked to 15 deaths and another to just one. It's tempting to call the first fifteen times as dangerous. But the count leaves out how many people took part. If far more people do the first activity, it can rack up more deaths while being the safer of the two for any one person who does it.
A death count is a total. What you actually want is a chance: the risk for one person, one go. That's what the micromort is built to hold. It also gives risks that arrive in different packaging (deaths a year, a chance per trip, a share of the people who try it) one common scale. A common currency, in other words, which tells you what each thing costs, not whether it's worth paying.
The unit works best for acute risks, the kind that kill straight away if they kill at all. Risks that act slowly, wearing down life expectancy over years, are a different shape of thing, and the micromort fits them less neatly.
Why a million
Tiny probabilities are awkward to say out loud. Written as a decimal, a one-in-a-million chance is a nought, a point, five more noughts and then a one. Easy to miscount. Hard to feel. Two such numbers side by side look almost identical even when one is several times the other.
A million moves the decimal point out of the way. Once the unit is a millionth, everyday small risks become whole numbers: one, five, eight. You can hold them in your head and tell at a glance that eight is bigger than one. Same risk, friendlier number, the way a pencil is easier to describe in centimetres than in fractions of a kilometre.
The paper that proposed the unit offered one example to put things in perspective: in the society it describes, people face about 270 micromorts a year from interactions with motor vehicles. A 270-in-a-million chance of dying from that cause over a year. The unit doesn't tell you whether that's high or low. It just makes it sayable, and comparable with everything else on the same scale.
The tempting step: just add them up
Lengths add. Walk one metre, then another, and you've walked two.
Micromorts, for small risks, behave in much the same way. Take two separate activities that each carry one micromort. Do both, and it's natural to say the combined risk is two. If you've ever assumed that doing something twice is about twice as risky, you've added micromorts in your head. For small risks, that answer is very close to right, and it's genuinely useful: a day, a trip or a year can be totalled from its parts with nothing fancier than addition.
But "very close" is doing real work there. Lengths add exactly. Chances of death don't.
Where adding breaks
Death, unlike distance, only happens once. You can walk a second metre after the first. You can't die a second time after the first.
The cleanest way to see the problem is an extreme case. Imagine an activity carrying half a million micromorts, a one-in-two chance of death. Do it twice and plain addition says half a million plus half a million: a million micromorts, certain death.
That's wrong. Each time, there's a one-in-two chance of surviving. Survive the first, and you face the same odds again. The chance of getting through both is a half of a half, a quarter. So the chance of dying is three in four: 750,000 micromorts, not a million.
What went wrong is double-counting. Plain addition counts the second risk in full, including the share that would only have applied to someone who'd already died the first time. The fix, in words: work out the chance of surviving each activity, multiply those survival chances together, and whatever's left is the chance of dying. Done that way, the combined risk always comes out below the simple total. How much below depends on the two risks multiplied together, then scaled back down by a million.
When both risks are small, that knock-off is tiny, far too small to matter. That's why adding works so well for everyday micromorts. Only when risks get large does the overlap grow big enough for plain addition to oversell the danger.
What the name buys you
Calling a distance a hundred metres doesn't make the walk any shorter, and calling a risk eight micromorts doesn't make it safer. What the name does is put small chances of death on one scale, in whole numbers you can compare at a glance. Keep them small and you can add them like lengths. Let them grow, and you have to remember the one thing a metre never has to: you only get to cross the finish line once.
SOURCES
Wikipedia, "Micromort." https://en.wikipedia.org/wiki/Micromort (accessed 24 Oct 2026)
Howard, R. A. (1989), "Microrisks for Medical Decision Analysis," International Journal of Technology Assessment in Health Care. https://doi.org/10.1017/s026646230000742x (accessed 24 Oct 2026)
NRICH, "Evidence and Headlines." https://nrich.maths.org/articles/evidence-and-headlines (accessed 24 Oct 2026)
Understanding Uncertainty, "Micromorts." http://understandinguncertainty.org/micromorts (accessed 24 Oct 2026, via web.archive.org)
Comments