Method, data and sources

How Our Death Clock Calculates Your Estimate

Our death clock uses the Gompertz-Makeham survival model. We tune it so each country and sex matches its life expectancy at birth from UN World Population Prospects 2024. It starts from your current age, adjusts risk for your answers using published hazard ratios, and adds up your chance of being alive in every future year. Everything below is public so you can check it.

By the Death Clock Calculator editorial team · Last reviewed September 2026 · How we calculate

The formula

The model describes the risk of dying at age x, called the force of mortality, as:

μ(x) = A + m · B · eb·x

  • A is a small, steady background risk that doesn’t change with age (accidents, some infections).
  • B · eb·x is the risk that grows with age. It doubles every fixed number of years.
  • m is your personal multiplier from BMI and the optional answers. With no extra risks, m = 1.

From this, the chance of still being alive at age x is:

S(x) = exp( −A·x − m·(B/b)·(eb·x − 1) )

The constants we use, and why

Model constants. B is solved separately for each of 402 country and sex groups.
ConstantValueWhy
b, rate of ageingln(2) ÷ 7 = 0.0990 per yearAdult death rates double roughly every 7 to 9 years in published studies. We use 7 years because, in our tests, it gave remaining life expectancy at older ages closer to official life tables for high-income countries than 8 or 9 years did.
A, background risk (men)0.0014 per yearMen have higher accident and injury deaths at young adult ages.
A, background risk (women)0.0007 per yearHalf the men’s value, reflecting lower injury deaths.
B, level of riskSolved for each country and sexFound by bisection so the curve’s life expectancy at birth equals the UN figure to within 0.01 years.
Examples of the solved level constant B (per year) for four countries.
CountryB, menB, women
United States1.753e-51.313e-5
India3.534e-53.120e-5
Pakistan6.250e-54.471e-5
United Kingdom1.327e-51.138e-5

Step by step

  1. Baseline. Your country and sex select a UN life expectancy at birth. We solve B so the curve matches it.
  2. Your age. Your date of birth gives your exact age today, x.
  3. Your multiplier. BMI and optional answers are combined into m, by multiplying their hazard ratios. We cap m between 0.55 and 5.
  4. Remaining years. We add up S(x + t) ÷ S(x) for every future step t of 0.05 years, up to age 125, with the trapezoid rule. This is your remaining life expectancy, given that you have reached age x.
  5. Estimate. Age at death = x + remaining years. The projected date is today plus the remaining years.
  6. Middle half. The ages at which your conditional survival falls to 75% and 25%.
  7. What moves your estimate. We re-run steps 3 to 5 with one answer changed at a time, and then with the changes you pick together.

How each answer changes risk

Hazard multipliers applied to the age-related part of risk. “Prefer not to say” always equals 1.00.
InputValues usedBasis
BMIUnder 18.5: 1.51 · 18.5 to 19.9: 1.13 · 20 to 24.9: 1.00 · 25 to 27.4: 1.07 · 27.5 to 29.9: 1.20 · 30 to 34.9: 1.45 · 35 to 39.9: 1.94 · 40+: 2.76Hazard ratios as published by the Global BMI Mortality Collaboration (Lancet, 2016)
Smoking nowUnder 10 years: 1.9 · 10 to 19: 2.3 · 20 to 29: 2.7 · 30+: 3.0Calibrated so lifelong smoking costs about 10 years, as in Doll and Peto (BMJ, 2004)
Former smokerShare of the smoking excess that remains: quit by 30: 0% · 40: 6% · 50: 28% · 60: 50% · 70: 82% (straight lines in between)Calibrated to the years gained by quitting at 30, 40, 50 and 60 in Doll and Peto
AlcoholNever or rarely: 1.00 · 2 to 4 times a month: 1.02 · 2 to 3 times a week: 1.06 · most days: 1.20 · several a day: 1.45Our weights. Direction and shape from GBD 2016 Alcohol Collaborators (Lancet, 2018)
ActivityMostly sitting: 1.25 · some: 1.07 · WHO target: 0.90 · above target: 0.82Our weights, set so the gap from least to most active is about 4 years, close to Moore and colleagues (PLOS Medicine, 2012)
DietProcessed: 1.15 · typical Western: 1.05 · balanced: 1.00 · Mediterranean: 0.90 · whole food and plants: 0.88Our weights. Kept well below the gains modelled by Fadnes and colleagues (PLOS Medicine, 2022), with direction from PREDIMED (NEJM, 2018)
OutlookVery hopeful: 0.93 · mostly positive: 0.97 · in between: 1.00 · often down or worried: 1.08Our weights, small on purpose. Direction from Lee and colleagues (PNAS, 2019) and Alimujiang and colleagues (JAMA Network Open, 2019)

Where a study reports a hazard ratio, we use it directly and let the survival curve turn it into years. Where it reports only a direction, we use small weights and say so here. Multiplying the effects assumes they are independent. In real life, risks overlap, so a long list of changes is an approximation.

The data

Every country figure is life expectancy at birth for men and for women from the UN World Population Prospects 2024, medium variant, estimates for 2023. We include all 201 countries and territories with at least 90,000 people. One release and one year for every country keeps the comparison fair. The full table is on life expectancy by country.

Cause-of-death shares on how will I die come from the WHO Global Health Estimates 2021. The world death clock uses the UN’s 2026 projection of world deaths and births.

How we test the model

  • For 12 checks (6 countries, both sexes, including the United States, India, Pakistan, Nigeria, Japan and Hong Kong), the curve returns the UN life expectancy at birth within 0.01 years. Our pass mark is 0.1 years.
  • For a fixed person, the expected age at death rises at every older attained age, as it should.
  • The smoking settings reproduce years lost and years gained within about 1 year of Doll and Peto.

Known limits

  1. The model sees an average person of your age, sex and country. It can’t see genes, illness, income or luck.
  2. Gompertz-Makeham has no special term for infant deaths. In countries with high child mortality, the curve can look slightly pessimistic for healthy adults.
  3. In the longest-lived countries, the model can overstate the chance of reaching 100.
  4. Lifestyle effects are multiplied as if independent. Real risks overlap.
  5. Death rates are from today. Future medical progress is not included, which likely makes estimates for young people too low.
  6. National averages hide large gaps by income and region.

Privacy

All maths runs in your browser. Your answers are never sent to our server, never stored, and never included in analytics. See our privacy policy.

Sources

  1. United Nations, DESA, Population Division. World Population Prospects 2024. population.un.org/wpp
  2. World Health Organization. Global Health Estimates 2021. who.int
  3. Gompertz B. Philosophical Transactions of the Royal Society, 1825; 115: 513-583.
  4. Makeham WM. Journal of the Institute of Actuaries, 1860; 8: 301-310.
  5. Global BMI Mortality Collaboration. The Lancet, 2016; 388: 776-786. doi
  6. Doll R, Peto R, et al. BMJ, 2004; 328: 1519. doi
  7. GBD 2016 Alcohol Collaborators. The Lancet, 2018; 392: 1015-1035. doi
  8. Moore SC, et al. PLOS Medicine, 2012; 9(11): e1001335. doi
  9. Estruch R, et al. NEJM, 2018; 378: e34. doi
  10. Fadnes LT, et al. PLOS Medicine, 2022; 19(2): e1003889. doi
  11. Lee LO, et al. PNAS, 2019; 116(37): 18357-18362. doi
  12. Alimujiang A, et al. JAMA Network Open, 2019; 2(5): e194270. doi

Change log

  • September 2026: Launch. UN World Population Prospects 2024 (2023 estimates) for 201 countries and territories. WHO Global Health Estimates 2021 for causes of death. Doubling period set to 7 years.

Found a mistake? Tell us through our contact page. We check every report. Back to the death clock calculator.

Educational estimates from public life tables. Not medical, financial or insurance advice.
© 2026 Death Clock Calculator. Free, no sign-up, funded by ads only.