Before the count

Nobody counted the world before the twentieth century. Scholars work it out from tax lists, harvests, graves and the size of cities, and they disagree. For the year 1 AD their answers run from 170 million to 400 million people: the highest is 2.4 times the lowest.

Biraben (1980)Durand (1974)Haub (1995)McEvedy & Jones (1978)Thomlinson (1975)United Nations (1973)United Nations (1999)US Census Bureauthe whole spread

How to read this: time runs left to right, from 10000 BC to 1950 AD. Higher up means more people, and each rule up is ten times more. Every short dash is one scholar's estimate, in that scholar's own colour (the key names them all); a thin line joins the two ends of an estimate given as a range, and the faint bar behind them covers the whole spread for that moment. The dates are not evenly spaced: the early gaps are thousands of years, the later ones a hundred.

The estimates are never averaged here. Where they disagree, the disagreement is the finding: for most of human history we do not know how many people were alive to better than a factor of two. By 1950 AD the answers had closed to 2.4 billion–2.6 billion, and from 1950 the United Nations counts every year, which is where the rest of this site begins.

Asia has held most of the world's people for as far back as anyone can estimate: about 72 of every 100 two thousand years ago, and 63 of every 100 in 1940.

How to read this: in each block every square is 1 of every 100 people alive at that moment, and the regions are listed with the largest first. Each region keeps its own colour in every block. The number on the right is that region's share, out of 100.

These are modelled reconstructions, not counts, and the further back they go the rougher they are. The regions are the ones the reconstruction uses, which is why Central America is counted with North America here while the UN's groups put it with Latin America.

In 1800 no country's babies could expect 43 years of life on average. The longest-lived was Iceland, at 42.9 years. In 2023 even the country with the shortest lives, Nigeria at 54.5 years, was above it. The longest, Hong Kong, reached 85.5.

Countries in a square: 1–5859–116117–174175–232

How to read this: left to right is time, in five-year slices. Bottom to top is how long a baby born then could expect to live, in two-year bands. The stronger a square's colour, the more countries sat in that band in those years, so the strong path climbing the picture is the whole world of countries moving up together. 186 countries are followed the whole way.

Life expectancy at birth is not a lifespan. When many babies died, the average was dragged far below the age most adults reached, so a figure of 30 never meant that people were old at 30. Dips are real: a war or a famine shows up as a stray coloured square low on the picture. Before 1950 these are Gapminder's reconstructions; from 1950 they are the UN's own figures.

England's parish registers make one of the longest records of births and deaths anywhere. In 1541, out of every 1,000 people, about 39.8 babies were born during the year and 29.4 people died. In 2015 it was 12.1 born and 9.2 died. Both fell by two thirds: the deaths first, the births a lifetime later.

borndiedthe gap between them

How to read this: squares in the strong colour are births and squares in the weaker colour are deaths, both counted per 1,000 people living, one pair per point in the record. The faint bar between them is how fast the population grew that year: the wider the bar, the faster. Where a death square sits above a birth square, more people died than were born.

The worst point in the whole record is 1681, when 35.4 of every 1,000 people died, more than were born. Demographers call this whole shape the transition, and every country on this site is somewhere along it: deaths fall first, births follow a generation or two later, and the population grows fast in between.

Before 1862 the record gives one point every five years, reconstructed from parish registers by Wrigley and Schofield; after it, the registration figures. Those reconstructions are published in a book we may show and credit but not hand on, so this is the one figure on the site with no table and no file behind it.

Every week for nearly three centuries, London's parish clerks printed how many people had died and of what. In the worst year, 1796, smallpox killed 3,548 Londoners, 18.39 of every 100 deaths in the city. By 1902 the same disease was 1.59 of every 100.

How to read this: one column per recorded year. The taller the column, the bigger a share of all the city's deaths that year were smallpox. Blank stretches are years the record is missing: 241 of the 274 years are here.

The worst year was also the year Edward Jenner first tried a vaccine. It took another lifetime, and a law making vaccination compulsory in 1853, before smallpox stopped being one of the ordinary ways to die in a big city. In 1980 it was declared gone from the world — the only human disease that has been.

In 1915, of every 1,000 babies under one year old in England and Wales, 109.7 died within the year. In 2015 it was 3.9, about one twenty-eighth as many. At every older age the fall is smaller, and for the very old it is smallest of all.

1. under 1109.73.9
2. 1 to 418.60.2
3. 5 to 93.80.1
4. 10 to 142.30.1
5. 15 to 193.50.2
6. 20 to 245.20.4
7. 25 to 295.40.4
8. 30 to 345.80.6
9. 35 to 397.20.9
10. 40 to 448.81.4
11. 45 to 4911.52
12. 50 to 5416.33.1
13. 55 to 5922.44.8
14. 60 to 6434.27.8
15. 65 to 6948.311.8
16. 70 to 7481.519.6
17. 75 to 79121.933.8
18. 80 or older212.2107.5

19152015

How to read this: one line per age group. The upper bar, in the stronger colour, is 1915 and the lower, in the weaker colour, is 2015; the longer the bar, the more people of that age died during the year, out of every 1,000 of them. The notches cut each bar into tens, and the number at the end is the rate itself.

None of this says people stopped dying. Everyone dies once, and the very old still die at high rates. What changed is when, and what people die of now, at each age takes that forward.

There is no single count of the world's people before 1950, so each question on this page uses the best published reconstruction for it, and says whose it is.

Sources: US Census Bureau · HYDE 3.2 · Gapminder · UN WPP 2024 · Wrigley & Schofield · London Bills of Mortality · UK ONS. Every source, its licence and how to cite it →

Verified 2026-09-30 · 73 canaries