Key takeaways
- For a UK male aged 65 in 2026, the ONS cohort tables put the median age at death at 86.1, with a 1 in 4 chance of reaching 92.1 and a 1 in 10 chance of reaching 96.4.
- The headline period figure for the same man is 18.7 further years, measured across 2022 to 2024. His 1 in 10 planning age sits 31.4 years out, a difference of 12.7 years.
- Females aged 65 years in the UK in 2024 could expect a further 22.7 years on the cohort tables, against 20.0 years for males, a gap of 2.7 years.
- 19.1% of girls and 12.0% of boys born in 2024 are projected to reach age 100, rising to 26.3% and 18.3% for the cohort born in 2049.
- Life expectancy at birth in the most deprived tenth of England was 73.2 years for males in 2022 to 2024, against 83.6 years in the least deprived tenth.
Life expectancy tables answer a narrower question than the one you are asking
You want to know how long the money has to last. The number you meet first is life expectancy at 65, and in the UK that is 18.7 further years for men and 21.2 for women, measured across 2022 to 2024 by the Office for National Statistics. It’s the wrong number for the job, twice over.
First, it’s a period figure. It takes the death rates actually recorded in 2022, 2023 and 2024 and applies them to a 65-year-old for the rest of their life, as though nobody’s odds ever improved again. ONS is direct about what that assumes: period life expectancies "use mortality rates from a single year (or group of years) and assume that those rates apply throughout the remainder of a person’s life."
Second, even the better measure is an average, and roughly half of any cohort outlives its average. A plan sized to the middle of a distribution runs out of road for everyone in the top half. So the useful question isn’t what the average is. It’s how far the tail runs, and that is what the percentiles below are for. For a man of 65 today the answer is about 11 years further than the average, and about 13 years further than the period figure he was probably quoted.
Period and cohort: two numbers, both called life expectancy
ONS publishes both, and the gap isn’t small. Its own worked example follows twins born in 2000, who turn 26 in 2026. "Period life expectancy for a female aged 26 years in 2026 is 58.0 additional years and for a male is 54.3 additional years." On the cohort basis the same two get 62.6 and 59.2. Same people, same day, 4.6 years apart for her and 4.9 for him.
One assumption separates them. The period table freezes today’s death rates forever. The cohort table walks the diagonal instead: a 26-year-old meets the rate projected for 27-year-olds in 2027, for 28-year-olds in 2028, and onward. If mortality keeps improving, the diagonal is the path a real person takes. ONS calls cohort life expectancy the more realistic measure for exactly that reason.
The size of that gap depends on how much future is left. ONS has measured it directly at birth: for cohorts born between 1950 and 2000, the difference between cohort and period life expectancy at birth in England and Wales "has been around 8 to 10 years". A newborn has a whole lifetime of projected improvement ahead of them. A 65-year-old has already banked most of the improvement they were ever going to get.
At 65 the gap narrows, because there is less future left in which to improve. Females aged 65 years in the UK in 2024 could expect a further 22.7 years on the cohort basis and males 20.0. Running the period calculation on the same 2024 file gives 21.4 and 19.0. That is 1.3 years for women and 1.0 for men, gained purely by letting the projection do what it was built to do.
What the life expectancy tables say at each age, in one table
This is the artifact. Each row is a person of that age in 2026, followed down the diagonal of the ONS 2024-based projection until the survivors run out. Life expectancy is the mean of that distribution. The median is the age at which half the cohort has died. The 1 in 4 and 1 in 10 columns are the ages at which a quarter and a tenth of the starting cohort are still alive.
| Age in 2026 | Male: life expectancy | Male: median age at death | Male: 1 in 4 reach | Male: 1 in 10 reach | Female: life expectancy | Female: median age at death | Female: 1 in 4 reach | Female: 1 in 10 reach |
|---|---|---|---|---|---|---|---|---|
| 50 | 34.0 | 86.3 | 92.8 | 97.3 | 37.3 | 89.8 | 95.5 | 99.4 |
| 55 | 29.1 | 86.0 | 92.5 | 96.9 | 32.3 | 89.5 | 95.1 | 99.1 |
| 60 | 24.5 | 85.9 | 92.2 | 96.6 | 27.5 | 89.3 | 94.8 | 98.8 |
| 65 | 20.2 | 86.1 | 92.1 | 96.4 | 22.9 | 89.2 | 94.6 | 98.6 |
| 70 | 16.0 | 86.5 | 92.0 | 96.2 | 18.4 | 89.2 | 94.4 | 98.3 |
| 75 | 12.3 | 87.3 | 92.3 | 96.3 | 14.2 | 89.6 | 94.5 | 98.2 |
| 80 | 9.1 | 88.7 | 93.0 | 96.6 | 10.5 | 90.4 | 94.8 | 98.3 |
Read across row 65. A man’s mean is 85.2 and his median is 86.1, close enough to be the same answer. The 1 in 4 age is 92.1 and the 1 in 10 age is 96.4. For a woman the four numbers are 87.9, 89.2, 94.6 and 98.6. The chart above plots those figures for both sexes at 65.
Notice what happens down the left-hand column. The median age at death for a man barely moves between 50 and 70, from 86.3 to 86.5, and then rises to 88.7 by age 80. Surviving is itself information. An 80-year-old has already passed through the ages that killed part of his cohort, so his remaining distribution shifts later. Life expectancy falls with age; the age you can expect to reach does not.
The average isn’t the planning number, and the gap runs to 11 years
Take the man of 65 again. His cohort life expectancy is 20.2 years, so his mean age at death is 85.2. His 1 in 10 age is 96.4, which is 31.4 years away. The distance between the two horizons is 11.2 years. A portfolio built to fund 20.2 years and one built to fund 31.4 years are not the same portfolio, and the difference is not a rounding error in a safe withdrawal rate calculation.
Against the period figure of 18.7 years the gap is wider still, at 12.7 years. For a woman of 65 the 1 in 10 horizon is 33.6 years. None of this says a tenth of people is the right place to plan. It says the choice of percentile is a decision, and quoting a single life expectancy number hides that the decision was ever made. That is also the hinge of the annuity vs drawdown question: an annuity transfers the tail to an insurer, and the tail is what the tables are measuring.
There is a third number worth knowing, because it is the one most people picture when they imagine dying of old age. In 2020 the most common age or "modal age" at death was 87.1 years for males and 89.3 years for females in England and Wales, while the median age at death was 81.8 and 85.5. Mean, median and mode land in three different places on a skewed distribution, and all three get called "life expectancy" in ordinary speech.
Two people in one household push the horizon further out
A joint plan doesn’t end at the first death. Of a man and a woman both aged 65 in 2026, the chance that at least one is alive at 90 is 64.3%, and at 95 it is 34.1%. Individually, the chance of reaching 90 is 33.6% for him and 46.3% for her. Two lives take the household figure to 64.3%, well above either of them.
That calculation treats the two lifetimes as independent, which they aren’t. Couples share housing, income, diet and habits, so their deaths are positively correlated, and any positive correlation lowers the joint figure below 64.3%. Treat it as a ceiling rather than an estimate. It’s still far above either individual number, which is the point.
How these percentiles were built, and how to check them
Everything above comes from one file: the ONS "Life tables, principal projection, UK" spreadsheet on the 2024-based projection. Its documented coverage is the probability of dying at each single year of age from 0 to 100, for every calendar year from 1951 to 2074, and the rate sheets inside it run further still. For a person aged 65 in 2026 the cohort table applies the rate for age 65 in 2026, then age 66 in 2027, and so on, multiplying survivors through. The median is where the survivor count crosses half. The 1 in 4 and 1 in 10 ages are where it crosses a quarter and a tenth, interpolated between whole ages.
The check that matters is whether the method reproduces what ONS itself publishes. Run it on the 2024 cohort and it returns 20.0 years at 65 for males and 22.7 for females, and 86.9 and 90.2 at birth. Those are the published headline figures to the decimal, including the tail above age 100 that the mean depends on. ONS publishes the survivor counts these columns are built from, but not the percentile ages themselves.
The strongest objection: a national table is not your table
The serious case against using any of this for an individual is that a UK-wide table averages over differences larger than everything the table measures. In 2022 to 2024, life expectancy at birth in the most deprived areas of England was 73.2 years for males and 78.3 for females, against 83.6 and 86.4 in the least deprived. The slope index of inequality across the ten deciles was 10.4 years for males and 8.0 for females. The healthy-life gap is wider again: 49.8 years of healthy life in the most deprived decile for men, against 69.2 in the least deprived.
Those spreads dwarf the 1.3-year period-versus-cohort correction and rival the 11.2-year spread between the mean and the 1 in 10 age. ONS says its own calculator "does not consider other factors that may affect life expectancy such as lifestyle or health conditions", which is a fair description of its limits.
The objection is right, and it isn’t quite subtractable. The deprivation figures are period life expectancy at birth for a geographic decile, not cohort survival from age 65 for a person. Part of any gap measured from birth is accumulated before 65 by definition, and someone who has reached 65 has already lived through that part of it. You can’t take 10.4 years off the 86.1 median and call it an answer. What you can say is that the table is a population centre of gravity, and an individual sits some unknown distance from it.
What the life expectancy tables cannot tell you
They’re projections, and ONS publishes the uncertainty rather than hiding it. Everything here uses the principal projection, which assumes long-term annual mortality improvement of 1.1% for ages 0 to 90. The variants use 1.9% and 0.5%. On the high variant, cohort life expectancy at birth for the 2024 cohort was 94.0 years for females and 91.2 for males. On the low variant it was 87.1 and 83.5. That’s a spread of 6.9 years for women and 7.7 for men on the same birth cohort, produced by one assumption about a rate nobody can observe yet.
The tables also stop. ONS states that its calculator "is available for ages 0 to 100 years" and that expectations of life at the oldest ages "are highly uncertain and have not been included". The percentiles above stay inside that range, but the mean does not, because computing a mean requires rates beyond 100.
And the recent direction of travel has not been upward. UK period life expectancy at birth in 2022 to 2024 was 83.0 years for females and 79.1 for males, and ONS notes that for males "it was still slightly below the 2017 to 2019 level". A projection calibrated on a stalling series is a projection with a live question inside it.
Set against that, the long record argues the other way. In England and Wales, the chance of a newborn surviving to age 90 was 3% for males and 10.3% for females in the 1900 birth cohort; for the 2020 cohort it is 54.3% and 63.8%. ONS makes the same point about the measure itself: had period life expectancy been used for the cohort born in 1950, their lifespan would have been understated by that 8 to 10 years. The frozen-rates measure has erred in one direction. That is a large part of why underspending in retirement is as common as it is.
What would change these numbers
If the improvement assumption moves. The difference between 0.5% and 1.9% a year is the difference between 83.5 and 91.2 years of cohort life expectancy for a boy born in 2024. The next projection base year is the thing to watch, not the headline that comes with it.
If the recent stall proves structural. Male period life expectancy at birth in 2022 to 2024 is still below its 2017 to 2019 level, and the principal projection nonetheless assumes improvement resumes at 1.1% a year. If that stall is a new level rather than a dip, the cohort tables are systematically long, and the correction runs the opposite way to the one this piece has been making.
If the statutory horizon shifts again. State Pension age is legislated to reach 67 between 2026 and 2028, and 68 between 2044 and 2046. Each rise lengthens the stretch a private pot has to cover on its own and shortens the years the state pension is in payment.
The single number to carry away is not a life expectancy at all. It’s the distance between the age you would plan to on the average, 85.2 for a man of 65, and the age a tenth of his cohort reaches, 96.4. Whatever the withdrawal method, that distance is the quantity being insured against, and it’s why the 25x rule and its variants are really arguments about which percentile to fund. LedgerTouch projects a portfolio against a horizon you set; the tables are how you choose the horizon.