Calculator output
Death Date Calculator: Find the Exact Date You Might Die
A death date calculator turns a survival distribution into a projected month and year, built from the Gompertz–Makeham survival function and conditioned on the age you have already reached.
On this page
What is a death date calculator, and how is it different from a life expectancy calculator?
A death date calculator converts a statistical survival age into a projected calendar date by adding the estimated remaining years to today. A life expectancy calculator stops at an age; a death date calculator takes the extra step to a month and year.
Both tools integrate the same survival function. A life expectancy calculator stops at the age; a death date calculator carries that age forward to a date. A life expectancy calculator reports an age at a chosen survival percentile and hands that number back. A death date calculator carries the arithmetic one step further, adding the remaining years to the current date so the output lands on a calendar rather than on a number line.
That step converts an age into a date a countdown can tick against. An age cannot tick; a date can. The conversion also fixes the resolution question that most readers arrive with, because a date implies a precision the underlying estimate does not carry.
Death Clock Calculator reports a month and a year. The engine computes no weekday and asserts none, so any tool naming a day of the week for your death is reporting precision its own model cannot supply. That gap between displayed precision and modelled precision is the single most common misreading of any death clock output.
The calculator returns a projected calendar date at month and year resolution, derived by adding the median remaining lifespan to the current date. Death Clock Calculator published methodology, checked August 2026
Understanding the conversion is the shortest route to why your date of birth is the minimum input the model needs.
Producing that date rests on one equation, the Gompertz–Makeham hazard.
What is the Gompertz–Makeham equation, and why does it power the death date output?
The Gompertz–Makeham hazard is μ(x) = A + B·e^(b·x). The constant A is the age-independent Makeham term, covering accidents and infections that do not rise with age. B·e^(b·x) is the exponential Gompertz term covering biological aging, and B is solved separately for every country and sex against published life tables.
Actuaries call the output of that equation the force of mortality: the instantaneous rate at which a person of age x is dying, rather than a probability spread across a year. Reading it as a rate is what makes the two terms separable, because one stays flat and the other does not.
The two terms, and where each came from
Benjamin Gompertz described the exponential term in 1825, after he observed that adult death rates climb geometrically rather than linearly. William Makeham added the constant in 1860 to absorb the deaths that arrive at a flat rate whatever a person’s age. That pairing has anchored actuarial hazard modeling for over a century. It separates aging from accident.
The constants this site uses
The age-independent Makeham term is fixed at 0.0016 per year for men and 0.0009 per year for women. The level parameter B is solved per country and sex, which forces each curve to reproduce that population’s own published life expectancy rather than a global average or a regional approximation.
A = 0.0016 per year for men and 0.0009 per year for women; B is solved separately for every country and sex so the curve reproduces that population’s published life expectancy. Makeham WM, Journal of the Institute of Actuaries 1860;8:301–310; Gompertz B, Philosophical Transactions of the Royal Society 1825;115:513–583
No goodness-of-fit statistic accompanies those constants, because no fit has been computed against the source tables. The full derivation sits in how the model calculates your death date.
The exponential half of the equation carries a rate, and that rate has a plain meaning.
Why does adult mortality roughly double every 8.5 years, and what does that mean for your death date?
Adult mortality risk roughly doubles every 8.5 years, which sets the exponential coefficient at b = ln(2)/8.5, about 0.0815 per year. A 50-year-old faces roughly twice the annual death probability of a 41-year-old, and that acceleration is why a projected death date shifts with each decade of age.
Doubling is what an exponential curve looks like from the inside. Expressing the coefficient as a doubling period rather than a decimal makes the consequence legible: risk at 60 is about double risk at 51, which is double again the risk at 43. Three decades multiply annual risk roughly eightfold. Nothing unusual has to happen.
Demographic literature reports doubling periods of roughly 8 to 9 years across human populations, measured in different countries and different centuries. That stability is itself the finding, and it is why one equation with two constants can describe a Japanese woman and a Nigerian man without a separate theory for each.
The pattern holds well across the adult range and loosens at both ends. In young adulthood the flat Makeham term dominates, so the curve is nearly level. At the oldest ages the observed rate appears to flatten below what the exponential predicts, which is an active question in demography rather than a settled correction.
Adult death risk doubles roughly every 8.5 years as calibrated here, giving b = ln(2)/8.5, about 0.0815 per year; published literature reports 8 to 9 years across populations. Gompertz B, Philosophical Transactions of the Royal Society 1825;115:513–583
Re-run the calculator at 60 rather than 50 and the date moves by more than the decade that passed, because surviving that decade also removed the people who did not.
Which raises the question of which life tables set that starting point.
What data sources calibrate the death date calculator for your country and sex?
Country and sex baselines come from WHO Global Health Observatory life tables and the United Nations World Population Prospects 2024 revision, covering 96 countries in total. United States mortality and cause-of-death context comes from CDC National Vital Statistics, and each source is named with its release.
Naming a release matters more than naming a body. A citation reading “WHO data” tells a reader nothing they can check, because the same organization publishes several series on different cycles. The current WHO life tables cover 2000 to 2021 and appeared in World Health Statistics 2024. World Health Statistics 2026 followed on 13 May 2026.
The UN population estimates behind those tables sit at their 2024 revision, with the successor postponed from 2026 to 2027. WHO and UN releases move. The citation practice does not.
Each source supplies a different part of the calculation. The life tables fix the shape of the survival curve for a country and sex. The population estimates supply the denominators those tables are computed against. The cause-of-death data supplies nothing to the date itself, and enters only where a reader asks what they are most likely to die of.
| Source | What it supplies | Current release |
|---|---|---|
| WHO Global Health Observatory | National life tables by country and sex | 2000 to 2021 series, in World Health Statistics 2024 |
| UN World Population Prospects | Population estimates under the tables | 2024 revision; next postponed to 2027 |
| CDC National Vital Statistics | United States mortality context | National Vital Statistics reports |
Ninety-six countries carry their own fitted curve, drawn from the national life tables and population estimates named above. WHO life tables, 2000–2021 series; United Nations, World Population Prospects 2024
Selecting a different country selects a different curve, which is the single largest lever on the output; the size of that effect sits in how national life tables change your estimate.
With a curve chosen, the rest is arithmetic in a fixed order.
How does the calculator convert a survival probability into a specific calendar date?
The model integrates the survival function forward from your current age rather than from birth, takes the median of that conditional distribution, and adds the remaining years to today’s date. The result carries roughly five to eight years of uncertainty either side.
The order below is fixed, and each step depends on the one before it. Conditioning before the curve is chosen would condition the wrong curve. Taking a mean rather than a median would pull the figure toward the long tail of very old ages, which no reader occupies.
- Select the curve. Country and sex choose one of the fitted national curves.
- Condition on attained age. The conditional survival probability is S(x)/S(a): survival to age x given survival to the age a already reached. Everyone who died before a leaves the reference group.
- Integrate forward. The curve is integrated from a rather than from zero, yielding remaining years instead of a lifetime average.
- Take the median. The figure reported is the median of that distribution, not its mean.
- Add to today. Remaining years are added to the current date. The output is a projected calendar date at month and year resolution.
Step two is where most competing tools diverge, and the divergence is not small. A calculator that skips it reports the same figure to a 30-year-old and a 70-year-old born in the same country, which is wrong for both.
Step five looks trivial and is not. Adding remaining years to today rather than to a birthday means the output moves every time the page is loaded, by exactly the time that has passed. That is correct behaviour for a countdown and surprising behaviour for anyone expecting a fixed date.
A realistic uncertainty range is roughly five to eight years either side of the date shown, and the number reported is a median rather than a prediction for any individual. Death Clock Calculator published methodology, checked August 2026
Half the people matching your profile die before that date and half after it. Reading its ticking form is covered in what a death countdown actually measures.
Step two, the conditioning on attained age, produces the largest single disagreement between death clocks.
Why does your projected death date change when you enter your current age instead of your birth year?
Reaching an age is evidence. Each year you survive, the people who died younger leave your reference group, so the conditional average moves up. This is why a death date built on attained age always falls later than one built on life expectancy at birth.
Life expectancy at birth averages across everyone born in a period, infants included. Those deaths pull the figure down permanently, which describes a population correctly and describes a living reader wrongly.
Why the gap widens with age
Conditional survival is at least as high as unconditional survival at every attained age above zero, and the difference compounds. A 30-year-old reading a birth-table figure is off by a year or two. A 75-year-old reading the same figure is off by considerably more, because seventy-five years of demonstrated survivorship have been discarded in the reading.
Why this is an identity rather than a modeling choice
The correction follows from the definition of conditional probability, not from any assumption about how people die. No calibration decision produces it and none can remove it, which is why a calculator that omits the step is wrong rather than merely different from this one.
United States life expectancy at birth reached an all-time high of 79.0 years in 2024.; a 70-year-old American reading that as an age at death is reading a figure that excludes the survivorship already demonstrated. WHO life tables, 2000–2021 series
The shift at 40, 60 and 70 is worked through in estimating your age of death from a birth date.
A different family of tools claims to skip this arithmetic entirely by learning mortality from data.
How does an actuarial death date differ from an AI mortality prediction?
They answer different questions. life2vec classifies whether someone in a Danish cohort dies within a fixed four-year window, reporting a Corrected Matthews correlation coefficient of 0.41. It produces no dates, is not publicly available, and its authors reject the widely quoted accuracy figure.
The popular version of this comparison is wrong, and correcting it is worth the space. Savcisens and colleagues trained a transformer on Danish register sequences and tested one downstream task: given a life sequence, could the model identify people aged 30 to 55 who would die within the following four years. That is binary classification with a fixed window, and it yields no date at all.
The metric matters as much as the task. Accuracy is a poor measure where the outcome is rare, because a model predicting that everyone survives scores highly while being useless. The Corrected Matthews correlation coefficient corrects for that imbalance, which is why the paper reports it and why a percentage lifted from press coverage misrepresents the result.
| Attribute | Gompertz–Makeham estimate | life2vec |
|---|---|---|
| Question answered | Years remaining, given attained age | Death within four years, yes or no |
| Output | A calendar date with a range | A classification, no date |
| Headline metric | None published; no fit computed | Median C-MCC 0.41 |
| Inspectable | Equation and constants published | Held under register privacy rules |
| Available to use | Free, in the browser | Not deployed anywhere |
life2vec reports a median Corrected Matthews correlation coefficient of 0.41, 95% CI 0.40 to 0.42, outperforming its baselines by 11 percent; the “78% accuracy” figure repeated across the web comes from press coverage rather than the paper, and the authors have published a correction saying so. Savcisens G, Eliassi-Rad T, Hansen LK, Mortensen LH, Lilleholt L, Rogers A, Zettler I, Lehmann S. Using sequences of life-events to predict human lives. Nature Computational Science 2024;4(1):43–56
One consequence follows immediately. Any site advertising a death prediction powered by that model is not running it, and its authors state plainly that it is connected to no service currently claiming otherwise. The registers it was trained on are not available outside Denmark, and access is governed rather than purchased.
Neither the survival model nor the classifier escapes the harder problem.
What are the known limits of a death date calculator, and when should you not trust the output?
No accuracy percentage appears on this site, because none has been computed against the source life tables. The model treats you as an average member of a category, uses period life tables rather than cohort projections, and cannot see anything specific to you.
What the model cannot see
It holds no information about your genes, your medical history, an undiagnosed condition or an accident. A period life table freezes mortality as measured now, so it does not model medicine improving across the decades it projects into.
What the underlying data cannot support
A further limit sits beneath the model. WHO reports that only about one third of countries meet its standards for high-quality mortality data, and that of roughly 61 million deaths worldwide in 2023, only about a third carried cause-of-death information. Baselines for countries with weak civil registration are modeled rather than counted.
Only about one third of countries meet WHO standards for high-quality mortality data, and of roughly 61 million deaths worldwide in 2023, about one third were reported with cause-of-death information. World Health Organization, World Health Statistics 2026, published 13 May 2026
Publishing a goodness-of-fit statistic would mean fitting this model against the source tables, measuring deviation at every adult age, and printing the result whatever it showed. That work has not been done, so the number is absent rather than estimated. The honest position is set out in whether a death clock can be accurate at all.
A projected death date is worth what its method is worth, and this method is written down. Read the date as what it is: the middle of a distribution drawn from national life tables, conditioned on the years already survived, carrying roughly five to eight years either side. Change one input and watch it move, because the movement is the part with practical meaning. The figure alone resolves nothing. What it can do is rank the things still under your control against each other, in years, using effect sizes drawn from named studies rather than from opinion. That ranking holds steady even where the date does not.
FAQs
What is a death date calculator?
It estimates a probable age at death from a survival model, then converts that age into a projected calendar date by adding the remaining years to today, at month and year resolution.
How accurate is a death date calculator?
No accuracy figure is published here, because no fit has been computed against the source life tables. The calculator displays roughly five to eight years of uncertainty either side.
Can you actually predict the date you will die?
No model predicts an individual death. The calculator reports the median of a distribution, and nothing in the arithmetic identifies which half of it contains any particular reader.
What information do you need to calculate your death date?
Date of birth is the only input the calculation cannot proceed without, because it supplies attained age. Sex and country select the curve; lifestyle answers refine the result.
What is the difference between a death date and a life expectancy?
Life expectancy is an age; a death date is a calendar date. The death date calculator carries the age forward by adding the remaining years to today.
How does a death clock turn an age into a calendar date?
It integrates the survival function forward from attained age, takes the median of that conditional distribution, and adds the remaining years to the current date. No weekday is computed.
Is my death date the same as my life expectancy date?
Not if the life expectancy figure is measured at birth. A birth-table figure includes everyone who died young, so it understates any living reader, and the gap widens with age.
What model does a death date calculator use?
Death Clock Calculator uses the Gompertz–Makeham hazard function, μ(x) = A + B·e^(b·x). The constant A covers background risk; the exponential term covers biological aging.
Why does my death date change when I enter my current age?
Because surviving to an age is evidence. Conditioning on attained age removes everyone who died younger from the reference group, which raises the conditional average.
What does the Gompertz law say about when I will die?
It describes how the force of mortality rises, not when any individual dies. Adult death risk climbs exponentially, doubling roughly every 8.5 years as calibrated here.
Leave a Reply