Life expectancy for Japanese women reaches 90

Where things stand

The latest MHLW figure is in the 2025 Abridged Life Table (令和7年簡易生命表), published 24 July 2026. It gives female life expectancy at birth of 87.33 years (up from 87.13 in 2024). The pre-COVID peak was 87.71 in 2020 (Table 1; annual series, 参考資料2; Nikkei and Kyodo, 24 July 2026). That release came out before the 16 September 2026 opening date, and its observation period is calendar 2025, so it cannot count.

Which releases are eligible. The observation period has to end after 16 September 2026, which rules out the 2025 Complete Life Table due around 2027. That leaves five annual Abridged Life Tables, each published in late July (24–26 July in 2024, 2025 and 2026):

• the 2026 table (July 2027)
• the 2027 table (July 2028)
• the 2028 table (July 2029)
• the 2029 table (July 2030)
• the 2030 table (July 2031)

Nothing qualifying can appear before Q3 2027, and the last real chance is July 2031.

Reference class and calculations

I took the female age-0 series for 1980–2025 from 参考資料2 and computed:

• Average annual change: +0.19 years for 1980–2025 (standard deviation 0.24). The pace has slowed to +0.07 per year since 2010 and +0.03 per year since 2015.
• Largest 5-year gain since 1980: +1.80 years (1981→1986), when the level was only about 79–81. The largest 5-year gain ending in 2000 or later is +1.75 (1995→2000).
• What YES needs: +2.67 years by the 2030 table, about 0.53 per year for five straight years. That is roughly three times the long-run average and beyond any 5-year gain recorded since 1980. Doing it by the 2026 table would take +2.67 in a single year.
• Mortality cut needed: I rebuilt the 2025 female table from MHLW's life-table spreadsheet; my rebuild gives 87.33. Reaching 90.0 needs death rates 27% lower at every age than in 2025, a fall of about 6.2% a year sustained over 2025–2030. Death rates in older age groups have historically improved by roughly 1–2% a year.
• Simple trend simulation: I treated each year's change as a random step around a trend. With the long-run trend (+0.19 per year, spread 0.24), the chance of reaching 90 by 2030 is about 0.1%. With a +0.10 trend it is about 0.005%. Both overstate the risk, because big drops are usually followed by rebounds (for example, 2011 then 2012).
• Global ceiling: no population has ever recorded female life expectancy of 90. Hong Kong leads the world at a provisional 88.7 for 2025 (Centre for Health Protection).
• Recent deaths: the latest monthly vital statistics show no collapse in mortality; July 2026 deaths were up 2.9% on a year earlier (MHLW preliminary figures via wic-net, 25 September 2026). A realistic central estimate for the 2030 table is about 88.0–88.8.

Ways it could resolve YES, and the case against my estimate

The strongest case for YES is a transformative medical shock, for example AI-accelerated therapies that sharply reduce cancer, heart disease and age-related (老衰) deaths, cutting deaths at every age by more than a quarter by calendar 2030. Even under a fast-takeoff scenario, trials, approval and rollout across Japan's very old female population make that implausible within the window.

The other routes are:

• a methodology change or new successor series; no plausible change adds 2.7 years
• an erroneous publication; a headline figure like this would be caught quickly

None of these justifies more than a fraction of a percent.

Forecast

I modelled the chance of first resolving YES in each quarter (the hazard). Quarters with no release get a background 0.005% per quarter. The July release quarters get 0.01%, 0.02%, 0.03%, 0.05% and 0.1%, rising with time for the AI and medical-breakthrough tail. Compounding these gives a cumulative probability of about 0.29% by the end of 2031. I would comfortably take 99:1 odds against YES.

Question. By the last day of each quarter, has Japan's Ministry of Health, Labour and Welfare reported national life expectancy at birth for women of 90.0 years or more?

Definitions.

• Life expectancy at birth for women: the female period-life-table expectation of remaining life at age 0 (女・0歳の平均余命, also called 平均寿命), for the national Japanese population covered by MHLW's life tables. This series covers Japanese nationals in Japan, not all residents regardless of nationality.
• Report: an official MHLW statistical publication reporting an observed-period estimate in its annual Abridged Life Tables (簡易生命表), five-yearly Complete Life Tables (完全生命表), or a successor national life-table series measuring the same quantity. Estimates based on provisional vital statistics and official revisions count.
• 90.0 years or more: at least 90.00 years using the most precise value published for that estimate in the release's statistical tables; do not round a value below 90.00 up to one decimal place.

Resolves YES if a qualifying MHLW publication reports the defined statistic at or above the threshold. The observation period must end after the opening date, under the Observation period convention. No attribution to AI is required. A revision that first raises an eligible estimate to the threshold counts on the revision's publication date; a later downward revision does not reverse YES.

Resolves NO for a quarter if no qualifying publication exists by its end. Exclude projections of future life expectancy, cohort-life-expectancy forecasts, healthy life expectancy, mean age at death, remaining life expectancy at older ages, cause-deleted hypothetical life tables, and prefectural, municipal, or other subgroup estimates. Re-publications or revisions solely concerning observation periods ending on or before the opening date do not count.

Resolving source. MHLW's official national life-table releases only, a narrower source requirement than Public confirmation (A2). Use the female age-0 entry in the life table or Table 1, 主な年齢の平均余命, of the annual summary, or its equivalent in a Complete Life Table. The MHLW life-table results index identifies releases; the 2025 annual release and its Table 1 identify the series and table. Apply Resolution date (A3) and the Resolution window convention.