Countries With Declining Populations: Who Is Shrinking, and How Fast

By the populationclock.org team · Published October 22, 2026 · All figures: UN World Population Prospects 2024, medium variant

The short answer

Some 65 of the 237 countries and territories in the dataset are losing people in 2026, in the strict sense that the medium projection puts next 1 January below this one. Between them they hold about 27% of humanity. A slightly different count, the number that have already passed their all-time peak, gives 63.

Both numbers are new. In 1990 only 23 countries and territories were losing people, and they held about 1.9% of the world’s population. Decline has gone from a statistical curiosity to the normal condition of a quarter of the planet in about one working lifetime.

The fastest decline is in small places

Ranked by the share of population lost between 2024 and 2100, the top of the table is Hong Kong at about 72%, Puerto Rico at about 69%, Saint Pierre and Miquelon at about 64% and Jamaica at about 62%. seven of the ten steepest declines belong to places with fewer than 10 million people today, which is what percentage rankings do: they reward a small base. Our shrinking ranking has the full list.

Put a size floor on it and the picture sharpens into something recognisable. The steepest falls among countries above 10 million people are Ukraine at about 59%, South Korea at about 57%, Taiwan at about 57%, China at about 55% and Poland at about 50%. Ukraine is the outlier in that group for having peaked back in 1993, before most of the others had stopped growing.

Where the people actually disappear

Percentages and people are different questions, and the second has a much shorter answer. The 110 countries and territories that are smaller in 2100 than in 2024 lose about 1.21 billion between them, and China alone accounts for about 65% of that, falling from roughly 1.42 billion to 639 million. The five largest declines, China, Brazil, Japan, South Korea and Thailand, make up about 77% of all the population lost anywhere.

That concentration is why the subject is usually reported through East Asia. China, Japan and South Korea are the clearest cases: Japan peaked in 2010 and is projected at about 77.0 million by 2100 against 123 million now, and South Korea loses about 57% of its population over the century.

The map of decline today

Decline is regional before it is global. Fully 30 of the 50 European countries and territories are shrinking this year, which is why Europe is already past its peak as a region. In Asia 8 of 51 are, including China, Japan, Thailand, South Korea and Taiwan. The Caribbean is the quiet entry on the list: some 14 of the 50 countries and territories in Latin America and the Caribbean are losing people, among them Cuba, Uruguay and Puerto Rico, with Cuba projected to fall by about 49% this century.

Africa is the exception that explains the world total. Just 2 of 58 African countries and territories are shrinking, Mauritius and Saint Helena, and at the far end of the projection 43 African countries are still adding people in the final year of the horizon.

Decline is a schedule, not a club

The useful way to read this dataset is as a queue. Countries are not divided into shrinking and growing so much as arriving, one by one, at the far side of their own peak. Germany peaked in 2024; Thailand in 2022. On the medium projection the shrinking countries become a majority of all countries around 2059, at 119 of 237, and by 2099 the count reaches 159.

The population-weighted version of that crossing is stranger. The shrinking countries first hold a majority of the world’s people around 2062, at about 51.3%, and then the share drifts back below half, to about 47.8% by 2095. The reason is arithmetic rather than recovery: the countries in decline keep getting smaller, so their share of the world shrinks along with them, while the countries still growing get bigger. By the end of the horizon 78 countries and territories are still adding people, and they hold about 51.8% of the 2100 population.

Peaked, shrinking, and not the same thing

The two counts at the top of this piece differ for a reason worth knowing. Having peaked is permanent in this dataset; losing people in a given year is not. three of the 63 countries past their peak are not projected to lose population this particular year, and a handful of small territories dip and recover more than once before settling. Our peaked ranking lists the first count and the shrinking ranking the second, which is why the two pages do not contain quite the same countries.

None of this means the world is shrinking, which is the usual inference and still the wrong one. The countries losing people shed about 1.21 billion between 2024 and 2100 while the rest add enough to lift the world total by about 2.06 billion over the same period. The world peak arrives later, in the 2080s.

How solid are these numbers

Firmer for the countries already in decline than for the ones scheduled to join. A country that peaked decades ago, with few young adults left to have children, is not going to resume growth from births alone, and the record of these projections is that fertility keeps coming in lower than assumed rather than higher. The soft assumption is migration, which is the one lever that can hold a national total up for years at a time, and the one the UN has least to go on. Our methodology page sets out what the low and high variants do to figures like these.

Frequently asked questions

All figures are the UN World Population Prospects 2024 medium variant. Shrinking means the projected 1 January total falls from one year to the next; peak years and population counts are the dataset’s own. A projection is a conditional statement about what follows from a set of assumptions, not a forecast.