What life expectancy actually measures
It is the most quoted number in demography, and almost nobody who quotes it knows what it is.
Ask what life expectancy means and most people will say something like: how long a person born now will live, on average. It is a reasonable guess. It is also not what the number is. Life expectancy at birth is not a forecast for anyone. It is a snapshot of one period's death rates, dressed up as a lifetime.
A snapshot, not a forecast
The figure that gets published each year is a period life expectancy, and it is built like this. Take one calendar year. Measure the death rate at every age in that year — what share of newborns died, what share of one-year-olds, and so on, all the way up. Then imagine a hypothetical batch of 100,000 babies and march them through every one of those rates in turn, as if the conditions of that single year were frozen in place for a century. The average age at which the imaginary babies die is the life expectancy at birth.
Nobody lives that life. A child born this year will not face this year's death rate at 70 when they reach 70; they will face whatever the rate is in the 2090s. The period figure is a thermometer reading of mortality now, expressed in years so that it feels like a prediction. It is a very good thermometer. It is not a crystal ball.
The alternative — a cohort life expectancy — follows people who were actually born in the same year and records how long they really lived. It is the true answer to the question everyone is asking. It is also unavailable for anyone still alive, because you cannot average a cohort's lifespans until the last of them has died. For anyone born much after 1910 it has to be projected, which is to say guessed carefully.
Why it can fall two years and bounce straight back
The snapshot nature of the number explains something that confused a lot of people during the pandemic. In the United States, life expectancy at birth was 78.8 years in 2019. It fell to 77.0 in 2020 and 76.4 in 2021 — a drop of nearly two and a half years, and the biggest two-year fall since 1921–23. It recovered to 77.5 in 2022 and 78.4 in 2023.
Read as a forecast, that is nonsense: the lifespans of everyone alive did not shrink by two years and then grow again. Read as a snapshot, it is exactly right. 2020 and 2021 had unusually high death rates at most ages, so a hypothetical cohort forced to live under those rates forever would have died younger. When the rates came down, so did the penalty. A period life expectancy is a report on how deadly a year was, and 2020 was a deadly year.
The infant mortality trap
The same logic dismantles the most durable myth about the past: that people once "dropped dead at forty". In England and Wales in 1841, the first year with a proper life table, life expectancy at birth was about 40 for a boy and 42 for a girl. But a child who survived to their fifth birthday could expect to live into their mid-fifties, and roughly one in three of them reached seventy. What dragged the average down was not adults dying young. It was that something like one child in seven died before their first birthday, and every one of those deaths entered the average as a lifespan of a few weeks or months.
That is the mechanism behind the whole modern rise. Global life expectancy at birth was roughly 32 years around 1900, about 46 in 1950, and reached 73.3 in 2024. In the first half of that run, most of the gain came not from old people living longer but from children not dying; only later did the old start getting their share. Clean water, vaccination, antibiotics and enough to eat did more for the headline number than any advance in geriatric medicine.
| Year | Global life expectancy at birth | What changed |
|---|---|---|
| c. 1900 | about 32 | Roughly two in five children dead before age five |
| 1950 | about 46 | Sanitation, clean water and public health reaching the rich world |
| 2000 | about 66 | Child survival transforming Asia and Latin America |
| 2024 | 73.3 | Back above the pre-pandemic level; child survival still the biggest driver worldwide, older ages a growing share |
Your number is not the headline number
This is the part that matters whenever the figure gets applied to an individual. Life expectancy is conditional on age. Every year a person survives, they leave behind the risks they have already passed, and their expected age at death goes up. The headline figure belongs to a newborn. Nobody reading this is a newborn.
The national life tables for the United Kingdom make the point plainly. Life expectancy at birth is about 79 for men and 83 for women. But a 65-year-old man can expect roughly another 19 years, taking him to about 84, and a 65-year-old woman roughly another 21. At 85, the remaining expectation is still six or seven years, not the negative number a naive reading of the headline would give. The further someone has come, the further they are expected to go.
This is why the question "what is the life expectancy?" is almost always the wrong one, and "what is the remaining life expectancy at my age?" is almost always the right one. Statistical offices publish both. Headlines almost always print only the first.
Why the gap between countries is thirty years
The highest national life expectancies at birth are now above 85; the lowest are in the mid-50s. That gap of more than thirty years is not mainly a story about old age. It is the infant mortality trap again, now geographic rather than historical: in most of the countries at the bottom of the table, between a third and a half of all deaths happen in childhood, many from infections that are cheap to prevent or treat. Bring child mortality down and the headline number climbs fast, which is precisely what has happened across much of Asia and Africa in living memory.
Today women outlive men in every country — by about five years globally — though it was not always so. As recently as the 1970s, men lived longer than women in India, Pakistan and Bangladesh, and the female advantage only became universal in the last few decades. The gap has biological components and behavioural ones, and demographers still argue over the proportions.
What the Ledger's calculator does with all this
The "How Will You Demise?" calculator on the dashboard does not deal in years at all. It starts from a rough set of base odds for each cause of death, nudges them according to the vices you tick, and draws one cause at random from the result. It is a novelty, and it says so. It offers a manner of death, never a date — and the reason nothing on this site will offer a date is worth stating. A life expectancy is a period figure, a snapshot of this year's death rates. Applying it to a future that will have different ones is guesswork dressed as arithmetic, and any tool that hands you a date is doing exactly that, whether it admits it or not.
The honest use of life expectancy is comparative. It tells you whether this year was deadlier than last, whether one country's mortality is heavier than another's, whether a policy moved the needle. It is a superb instrument for those questions. It was never designed to answer the one people ask it.
None of this is medical advice, and none of it is a prediction about anyone. A remaining-life-expectancy figure is an average across millions of people. It says nothing about any one of them.
The figures above come from the UN World Population Prospects (global and national estimates from 1950 onwards; the c. 1900 figure is Riley's historical estimate as compiled by Our World in Data), the US National Center for Health Statistics (United States life tables), the UK Office for National Statistics (national and historical life tables for England and Wales), and the historical series compiled by Our World in Data. All are free. The UN, NCHS and ONS publish full remaining-life-expectancy tables by age, and Our World in Data charts the same figures at a handful of ages — the tables this article argues are the ones worth reading instead of the headline.