people on
km²
That land
people on
km²
people live in on .
Loading the map
Loading detail
Where We Live, and Where We Don’t
Each column is one grid cell. Height follows the square root of density, so a cell 100 times denser stands 10 times taller.
The darker patch on the hovered country is the circle. It is the land around your pointer that holds as many people as the capital, with the capital itself left out. Where people live close together the circle stays small; across empty land it can cover most of a country.
Cities here are urban areas measured from space, not city limits: Paris has 9.3 million people, the City of Paris 2.1 million.
A city is a GHSL Urban Centre: a continuous patch of dense built-up land with at least 50,000 people, drawn by the same rule everywhere. Cities that have grown together count as one, which is why Guangzhou, at 43 million, is the largest city in China: its Urban Centre takes in most of the Pearl River Delta. A capital too small for an Urban Centre gets a settlement cluster from the same GHSL data, and the readout says so. Capitals come from Wikidata; where a country has several, the map picks one, such as Pretoria or La Paz.
Population is the GHSL grid GHS-POP for 2025, with cells about 100 m across. The 2025 figures are a model estimate: census counts extrapolated to 2025 and scaled to the UN World Population Prospects 2022, then spread over the cells by building volume, which satellites mapped mostly up to 2018. A country’s total is the sum of its cells, and no official figures are mixed in.
Country outlines are the World Bank Official Boundaries. A disputed area goes to the state that administers it today, which says nothing about who should. Western Sahara, the Falkland Islands, South Georgia and the British Indian Ocean Territory are territories of their own, and Taiwan is a separate country, which the World Bank map does not show. Places under 100,000 people are drawn but do not react to the pointer.
Each country is drawn in an equal-area projection chosen for its shape (Albers, Lambert azimuthal or cylindrical), so areas are not distorted. Alaska and Hawaii are insets.
The 2025 numbers are projected from censuses taken mostly between 2010 and 2020 and from buildings mapped up to 2018. Fast-growing places, and places emptied by war or displacement since then, are the least reliable.
An independent study in 2025 (Láng-Ritter et al., Nature Communications) found that global population grids, this one included, undercount people in the countryside, often by a wide margin. Where few people live, the circle is probably drawn too large.
Where a capital’s point falls outside its built-up area, the map can pick the wrong settlement, as with Praia in Cape Verde.
Outlines are simplified. Coastal cells with less than a tenth of land have no column on the map, though their people are counted. Some small places carry odd names from the source data.