Three open datasets, one continuous series.
Where the numbers come from — HYDE 3.3, Gapminder v7, and UN WPP 2024 — their coverage, methods, and licenses.
Terms in this category
A historical reconstruction of global and country-level population back to 10,000 BCE, built from archaeology and land-use models.
HYDE (History Database of the Global Environment) is produced by the Netherlands Environmental Assessment Agency (PBL). It uses land-use change, agricultural spread, and archaeological evidence to infer plausible population distributions. Used here for data before 1800. License: CC BY 3.0.
An open dataset of long-run country population statistics assembled from historical censuses and official records, 1800–1949.
Gapminder is a Swedish non-profit organisation. Their population dataset (v7) stitches together national census results, colonial administrative records, and scholarly estimates for countries that lack complete official series. It is designed to bridge pre-modern reconstruction and modern UN data. License: CC BY 4.0.
The 2024 edition of the UN World Population Prospects — the primary source for population data from 1950 onward and projections to 2100.
Published every two years by the UN Department of Economic and Social Affairs (DESA), the WPP harmonises national census and survey data using a set of demographic models. It provides historical estimates (1950–present) and projections (to 2100) at country, regional, and global level. License: CC BY 3.0 IGO.
A population figure from the UN World Population Prospects based on census, survey, and vital registration data.
UN estimates are produced by fitting demographic models to all available national data for each country. They differ from the raw census numbers in that they adjust for known biases, undercount, and coverage errors. From 1950 onward, these are the figures shown as solid lines on the chart.
A Creative Commons copyright license that lets anyone reuse data for any purpose, as long as the original source is credited.
CC BY (Attribution) is the most permissive Creative Commons license. The "IGO" suffix (as in CC BY 3.0 IGO) is a variant for intergovernmental organisations. All three datasets we use carry some form of open license, which is why we can build on them without restrictions.
Common questions
Why do you use three different sources instead of one?
No single dataset covers all of human history with consistent quality. HYDE reconstructs the distant past from indirect evidence; Gapminder compiles the modern-but-pre-census era from historical records; and the UN WPP provides rigorous contemporary estimates and projections. Each covers what it does best.
Where exactly do the seams between datasets fall?
The HYDE-to-Gapminder seam is around 1800. Before 1800 we use HYDE 3.3; from 1800 to 1949 we use Gapminder v7. The Gapminder-to-UN seam is 1950, after which UN WPP 2024 takes over through to 2100. Both seam years are marked with dashed lines on every chart.
Can I download the raw data from this site?
You can export a CSV of the current chart view using the Export button in the explorer. For the full underlying datasets, download directly from the source: HYDE at pbl.nl, Gapminder at gapminder.org, and UN WPP at population.un.org.
How often does the UN update its population data?
The UN publishes a new edition of the World Population Prospects approximately every two years. Revisions can change historical figures significantly, especially for countries where new census data has become available. We track the edition year (currently 2024) on every page.
Are there other population datasets I should know about?
Yes. The World Bank uses a subset of UN WPP data. The Human Mortality Database (HMD) covers high-income countries in more depth. The Human Fertility Database (HFD) goes into detail on fertility trends. The Maddison Project covers population alongside GDP for very long-run economic history.