Japan Census Records Biggest Population Drop as Only Tokyo Grows

Population falls by 3.17 million in five years

Japan's population fell to 122,972,528 on October 1, 2025, a decline of 3,173,571 from the 2020 census and the largest decrease recorded in a census series that began in 1920. The final figures put the national decline at 2.5%, compared with 0.7% in the previous five year period.

The Ministry of Internal Affairs and Communications' census shows a country shrinking across almost its entire territory. Tokyo was the only one of Japan's 47 prefectures to gain residents. Seven prefectures that grew in the 2020 census, including several major urban areas, moved into decline.

At the same time, foreign residents increased to a record 3.84 million, while the number of Japanese citizens fell by about 4.27 million. That growth softened the national population loss, but was not large enough to reverse it. People aged 65 and older accounted for 29.4% of residents, while children under 15 made up 10.9%.

The final figures also differ from the preliminary count released in May 2026, which put the population slightly higher and showed growth in Okinawa as well as Tokyo. Distinguishing those releases matters: the later count changes both the national total and the picture of which prefectures are still expanding.

A faster decline across three censuses

Japan conducts its census every five years. The population first fell in the 2015 survey, declined again in 2020 and recorded its third consecutive decrease in 2025. The latest loss was the largest by both number of people and percentage.

The change from a 0.7% decline to 2.5% represents an increase of 1.8 percentage points. Using those rounded rates, the pace of contraction was about 3.6 times that of the preceding census period. The final population and reported decrease imply a 2020 total of 126,146,099.

The ministry attributed the preliminary decline to population aging and a widening natural decrease, meaning that deaths exceeded births by a growing margin. Natural decrease is distinct from migration: a country can gain residents from abroad and still shrink if the balance of births and deaths is sufficiently negative.

The census establishes how many people live in Japan and where they live. The figures discussed here do not provide a complete numerical breakdown of births, deaths and migration over the five years, so they should not be treated as a full accounting of each cause of the decline.

Why the preliminary and final totals differ

The May 29, 2026 preliminary release counted 123,049,524 residents, down 3,096,575 from 2020. The final tally was 76,996 lower, with a correspondingly larger population loss. Both releases rounded the national decline to 2.5%, which can conceal the difference when only the headline percentage is used.

The ministry's preliminary census dataset identifies the survey period as October 2025 and its publication time as May 29, 2026, at 08:30. It covers population by sex, prefecture and municipality, alongside household counts, population density and changes over five years. Its comparisons use readjusted 2020 figures.

The release sequence separates an early population count from more detailed results:

• October 1, 2025: the reference date for the census population.
• May 29, 2026: publication of preliminary population and household counts.
• September 29, 2026: the scheduled release of the basic population and household results, also the date reported for the final announcement.
• By December 2026: scheduled publication of population migration results.
• Through November 2027: further releases covering employment, commuting and detailed occupations.

The Statistics Bureau's census results page distinguishes the preliminary count from the basic population and household tabulation. It also lists a reference table estimating the foreign population on October 1, 2025. Those categories should not be treated as interchangeable.

Tokyo grows more slowly as other prefectures shrink

In the final count, Tokyo's population increased by 1.9%, compared with 3.9% in the 2020 survey. Its growth rate therefore fell by two percentage points and was less than half the previous rate. Even Japan's remaining area of population growth was expanding more slowly.

Saitama, Chiba, Kanagawa, Aichi, Shiga, Fukuoka and Okinawa all shifted from growth in 2020 to decline in 2025. The other 39 prefectures recorded larger population declines than in the preceding census period. The pattern extends beyond places that had already been losing residents: areas that previously offered some balance to the national decline also contracted.

Okinawa illustrates the difference between the releases. The preliminary count showed a gain of about 1,000 residents, making it one of two growing prefectures alongside Tokyo. The final account places Okinawa among the declining prefectures. No explanation for that revision is given in the reported figures.

The preliminary release also recorded a Tokyo increase of about 199,000 people and identified Hokkaido as having the largest numerical loss, at 239,000. Shizuoka followed with a decrease of 164,000 and Hyogo with 141,000. Those are preliminary figures, not confirmed final amounts, and should be read with that distinction intact.

Foreign residents offset part of the loss

The reported number of Japanese citizens was 119,131,935, or 96.9% of the population, down 4,266,973 from 2020. That was a decline of 3.5%, compared with 1.4% in the previous census. Foreign residents numbered 3,840,039, or 3.1%, after an increase of 1,093,525, equivalent to 39.8%.

Measured against the reported loss of Japanese citizens, the foreign resident increase offset about 25.6%. Put another way, the increase was roughly one person for every four Japanese citizens lost from the count. This comparison shows why a record foreign population and a record national decline can occur at the same time.

China remained the largest nationality group, representing 24.0% of foreign residents, or about 923,000 people. Vietnam accounted for 15.9%, or about 609,000. South and North Korea together accounted for 10.4%, or about 398,000.

The relative position of the Vietnamese and Korean populations changed between censuses. In 2020, South and North Korean residents represented 15.8% of the foreign population, ahead of Vietnam's 13.9%. By 2025, Vietnam's share had risen by two percentage points, while the combined Korean share had fallen by 5.4 points. A falling share does not, by itself, establish a falling number of residents, because the total foreign population grew.

There is a small unresolved discrepancy in the precise totals: the stated Japanese and foreign resident counts add up to 122,971,974, which is 554 below the headline population. The figures do not explain that difference. It is too small to change the broad trend, but prevents an exact reconciliation of the categories.

Older residents outnumber children by nearly three to one

The census counted 36.142 million people aged 65 or older, representing 29.4% of the population. Their share rose from 28.5% in 2020, an increase of 0.9 percentage points. Children under 15 numbered 13.349 million, or 10.9%.

Those counts mean there were about 2.7 residents aged 65 or older for every child under 15. The gap between the two groups was approximately 22.793 million people. These are comparisons of age groups, not a direct measure of how many people are working or financially supporting others.

The ministry said the share of children was the second lowest after South Korea when Japan was compared with the other Group of Seven nations, China and South Korea. That is a comparison within the specified group, not a claim about every country in the world.

The reported population comprised about 59.7 million men and 63.2 million women. These rounded figures indicate roughly 3.5 million more women than men, but are not precise enough to reproduce the national total. The age and sex breakdowns add detail to the population loss without establishing each person's employment, health or living arrangements.

Smaller households change the meaning of population loss

The later census account put Japan's household count at about 57.3 million, with an average of 2.1 members, down from 2.21 in the previous comparison. Yamagata had the largest average household size among prefectures, at 2.41 people, while Tokyo had the smallest, at 1.85.

The preliminary release gave a different set of household figures: 57,124,507 households, an increase of about 1.294 million from 2020, and an average size of 2.15 people, down from 2.26. It described the household count as the highest since comparable data became available in 1970. The ministry said the smaller average likely reflected a growing number of older people living alone.

The preliminary and later accounts therefore differ in both the household total and the average size, including the earlier comparison value. The published descriptions do not establish whether every average uses identical definitions or treatment of households. Those measures should not be combined into a single precise series without checking the underlying tables.

The distinction between residents and households matters for understanding the results. A falling population does not necessarily produce fewer households: more people living separately can increase the household count even as the number of residents declines. The preliminary figures show that combination directly, rather than supporting an assumption that housing demand must move in step with population.

The census also affects electoral boundaries

The ministry used the latest census results to calculate differences in voter representation in the Lower House. Fukuoka No. 2 had the largest number of eligible voters, while Ishikawa No. 3 had the fewest. The ratio was 2.275 to one.

This means that, under the reported calculation, Fukuoka No. 2 had 2.275 times as many eligible voters represented by one constituency seat as Ishikawa No. 3. Forty districts had at least twice the eligible voter population of Ishikawa No. 3. Population change can therefore affect representation as well as the size and composition of communities.

A government council is to discuss a new plan for constituency boundaries, with the aim of reducing the disparity to less than two to one. The figures do not establish which boundaries will change or when recommendations will be completed.

There will be no reallocation of seats among prefectures on the basis of this census because the 2025 survey was a simplified census between the more extensive surveys conducted every ten years. Changing district boundaries and changing the number of seats assigned to each prefecture are separate decisions.

Further releases will fill gaps in the picture

The Statistics Bureau's English release schedule lists population migration results by December 2026, labour force results by March 2027, workplace and schooling results by May 2027, and migration results linked to employment by June 2027. Detailed industry and occupation results are scheduled by November 2027.

Those tables can help distinguish where people live from where they work or study, and identify movement between places of residence. They will provide information that the national population total alone cannot supply. Results for smaller areas are scheduled to follow the relevant main releases.

The bureau also states that unknown values were filled where possible using CANCEIS, a statistical data editing and imputation system. Imputation assigns values to missing or unknown entries through statistical processing; it does not mean every entry was directly reported. Some unknown values could not be filled and remain unknown.

The official pages link to imputed 2025 results and a retroactive 2020 tabulation. Careful comparisons therefore require attention to release stage, definitions and treatment of missing information. The central finding is nevertheless consistent across the preliminary and final counts: Japan lost population faster than in the preceding census period, despite a substantial rise in foreign residents.

Key Points

• Japan's final 2025 census population was 122,972,528, down 3,173,571, or 2.5%, from 2020.
• The decline was the third consecutive census decrease and the largest recorded by number and percentage.
• Tokyo alone grew in the final count, and its growth slowed from 3.9% to 1.9%.
• Foreign residents reached about 3.84 million, offsetting roughly one quarter of the reported decline in Japanese citizens.
• People aged 65 or older represented 29.4% of residents; children under 15 represented 10.9%.
• The preliminary count was 76,996 higher than the final total and had also shown growth in Okinawa.
• A government council will discuss electoral boundary changes after the largest reported voter disparity reached 2.275 to one.
• Further census releases covering migration, employment and commuting are scheduled through November 2027.