Japan’s police keep a separate count of crime involving foreign nationals, and they have kept it since 1989. The series runs for 37 years without a break, split into two offence types, and it is public.
Three things are true of it at once.
In 2025 the police cleared 25,480 cases in that count. In 2005 they cleared 47,865. The 2025 figure is 53 per cent of the earlier one.
The count has also risen in each of the last three years, 2023, 2024 and 2025.
And over the most recent eleven years of the series, the number of persons cleared per 100,000 of the population the police use as a reference group fell from 682 to 366, because that population grew faster than the count did.
None of those three sentences contradicts the others. Which one you quote decides what the series appears to say.
What this count includes, and what it leaves out
The National Police Agency’s term is 来日外国人, which its own publications render as visiting foreigners. The definition is a subtraction: foreign nationals in Japan, minus permanent residents, minus spouses and children of permanent residents, minus special permanent residents. That last group is mainly long-settled Korean and Chinese families, many of them born in Japan.
So the count deliberately excludes the most settled foreign residents. What it includes is broader than the name suggests: short-term visitors and people with no valid residence status can both appear in it.
That matters for the arithmetic later on. The residence statistics used here as a denominator count people holding a mid-to-long-term status on a single day at the end of June. The numerator can include people the denominator never counted. When the two are divided, the result is a ratio to a mid-year population count. It is not a crime rate, and this article does not call it one.
One further distinction runs through everything below. Japanese police statistics count 検挙, which is not arrest. It covers cases referred to prosecutors on paper, without anyone being taken into custody, as well as cases where someone is. The English used here is cases cleared and persons cleared. A cleared case is a record of police activity. It is not a crime proven, and it is not a conviction.
The long series: cases cleared, 1989 to 2025
| Year | Total cases cleared | Penal code | Special law |
|---|---|---|---|
| 1989 | 5,765 | 3,572 | 2,193 |
| 1995 | 24,374 | 17,213 | 7,161 |
| 2000 | 30,971 | 22,947 | 8,024 |
| 2005 | 47,865 | 33,037 | 14,828 |
| 2010 | 19,809 | 14,025 | 5,784 |
| 2015 | 14,267 | 9,417 | 4,850 |
| 2016 | 14,133 | 9,043 | 5,090 |
| 2020 | 17,865 | 9,512 | 8,353 |
| 2022 | 14,662 | 8,548 | 6,114 |
| 2023 | 18,088 | 10,040 | 8,048 |
| 2024 | 21,794 | 13,405 | 8,389 |
| 2025 | 25,480 | 17,614 | 7,866 |
The shape is a hill. The count climbed through the 1990s, peaked in 2005 at 47,865, fell for most of the following decade to a low of 14,133 in 2016, and has climbed again since.
2025 sits at roughly half the peak and roughly four and a half times the 1989 figure.
Persons cleared peaked in a different year
Cases and people are not the same measure, and they do not peak together.
| Year | Total persons cleared | Penal code | Special law |
|---|---|---|---|
| 1989 | 4,618 | 2,989 | 1,629 |
| 2004 | 21,842 | 8,898 | 12,944 |
| 2005 | 21,178 | 8,505 | 12,673 |
| 2010 | 11,858 | 6,710 | 5,148 |
| 2015 | 10,042 | 6,187 | 3,855 |
| 2020 | 11,756 | 5,634 | 6,122 |
| 2022 | 9,548 | 5,014 | 4,534 |
| 2023 | 11,534 | 5,735 | 5,799 |
| 2024 | 12,170 | 6,368 | 5,802 |
| 2025 | 12,777 | 7,333 | 5,444 |
Persons cleared peaked in 2004, a year before cases. In 2025 the figure is 12,777, or 58 per cent of that peak.
The gap between the two measures is worth holding on to. In 2025 cases rose 16.9 per cent on the year while persons rose 5.0 per cent. A small number of people can account for a large number of cases, and a change in one measure does not have to appear in the other.
Most of the decline was in special-law offences
This is visible in persons cleared, and it is not visible in cases. The distinction matters enough to state before the numbers.
Penal code offences are ordinary crime: theft, assault, fraud. Special-law offences are everything else on the statute book, and for this population the group includes both immigration control violations, which only a foreign national can commit, and drug offences.
| Persons cleared | 2004 peak | 2025 | 2025 as % of peak |
|---|---|---|---|
| Penal code offences | 8,898 | 7,333 | 82% |
| Special-law offences | 12,944 | 5,444 | 42% |
| Total | 21,842 | 12,777 | 58% |
Persons cleared fell by 9,065 between 2004 and 2025. About 83 per cent of that fall is in the special-law column. Persons cleared for ordinary crime are at 82 per cent of the earlier figure, which over 21 years is closer to flat than to falling — though in cases, as the table above shows, the same category more than halved.
Measured in cases the split disappears entirely. Penal code cases fell to 53 per cent of their 2005 level; special-law cases fell to 53 per cent of theirs. Identical.
So a single sentence saying that crime involving foreign nationals in Japan has halved is describing, in persons, the special-law column more than the penal-code column — and, in cases, both equally. The measure has to be named for the claim to mean anything.
The last three years
The National Police Agency’s own summary of the series is that totals were broadly flat from 2012 to 2020, fell for two years, and have risen in each of 2023, 2024 and 2025.
| 2021 | 2022 | 2023 | 2024 | 2025 | |
|---|---|---|---|---|---|
| Cases cleared | 15,893 | 14,662 | 18,088 | 21,794 | 25,480 |
| Persons cleared | 10,677 | 9,548 | 11,534 | 12,170 | 12,777 |
The starting point of that climb needs stating plainly, because it changes how large the climb looks. 2021 and 2022 are the years Japan’s borders were largely closed. The 2022 figure of 9,548 persons cleared is the lowest in the series since 2012. The rise begins from a closed-border low and coincides with the reopening; this series on its own cannot establish the cause.
For the 2024 increase the agency does give specific attributions: thefts by Vietnamese and Cambodian nationals in the penal code column, drug offences involving Filipino and Thai nationals in the special-law column. It offers no equivalent attribution for 2025 in the confirmed edition, and the special-law column fell that year, by 6.2 per cent in persons.
Against the population
The reference population is published as a note in the police white papers, and it can be reconstructed from immigration statistics for the years the white papers leave it out. For 2020 both methods give the same answer, 1,799,004 against the white paper’s about 1.799 million, which is the check that the reconstruction is doing what the police did.
| Year (end of June) | Reference population | Persons cleared | Per 100,000 | Penal code only | Per 100,000 |
|---|---|---|---|---|---|
| 2015 | 1,472,428 | 10,042 | 682 | 6,187 | 420 |
| 2016 | 1,677,441 | 10,109 | 603 | 6,097 | 364 |
| 2017 | 1,896,628 | 10,828 | 571 | 6,113 | 322 |
| 2018 | 2,092,296 | 11,082 | 530 | 5,844 | 279 |
| 2019 | 2,322,698 | 11,655 | 502 | 5,563 | 240 |
| 2020 | 1,799,004 | 11,756 | 654 | 5,634 | 313 |
| 2021 | ~1,698,000 | 10,677 | 629 | 5,573 | 328 |
| 2022 | ~1,838,000 | 9,548 | 520 | 5,014 | 273 |
| 2023 | ~2,558,000 | 11,534 | 451 | 5,735 | 224 |
| 2024 | ~3,120,000 | 12,170 | 390 | 6,368 | 204 |
| 2025 | ~3,490,000 | 12,777 | 366 | 7,333 | 210 |
Read down the columns and the eleven years say three things at once, all of them true:
- The counts rose. Persons cleared went from 10,042 to 12,777, up 27 per cent. For penal code offences alone the rise is from 6,187 to 7,333, up about 19 per cent.
- The recent direction is up. Increases in 2023, 2024 and 2025.
- The ratio to the reference population fell by almost half. 682 per 100,000 in 2015, 366 in 2025, down 46 per cent. For penal code offences alone it fell from 420 to 210, exactly half.
The middle years carry a warning. The ratio jumped in 2020 and stayed elevated in 2021, mainly because the denominator shrank: the reference population fell from 2.32 million in mid-2019 to 1.70 million in mid-2021 as the border closed. A ratio can move because of its denominator, and here it did.
What these numbers cannot tell you
- Cleared is not committed. These are cases police cleared and people police sent forward. Crimes nobody reported, and crimes nobody solved, are not in here. Neither are convictions.
- The numerator and denominator are not the same group. Short-term visitors and people without valid status can be cleared but are not in the residence count. Visitor numbers over this period moved far more sharply than resident numbers.
- A day is not a year. The population figures are a snapshot of 30 June. The clearance figures cover twelve months.
- Police activity changes. Enforcement priorities, staffing and the existence of dedicated units all move these counts, and none of that appears in the tables.
- Victims have no nationality here. The series records the nationality of people cleared. It does not record who was harmed, so it cannot say who bears the cost of any of this.
The wider picture
Two questions sit next to this one and are answered elsewhere.
The first is the share: what proportion of all crime in Japan involves foreign nationals. That requires dividing these counts into national totals, which brings its own definitional problems, and we cover it in Is Crime Rising in Japan?.
The second is why perceptions and statistics diverge. The same police agency that publishes these tables also publishes survey work finding that 79.7 per cent of respondents thought public safety had worsened, while 60.3 per cent described current safety as good, and that the most cited reason for the sense of worsening was television coverage at 76.4 per cent, against personal experience of crime at 14.2 per cent. That finding is about crime reporting in general rather than about foreign nationals specifically, and we treat it in the same article.
What the 37-year series supports on its own is narrower, and it is this: the count is around half its mid-2000s peak; in persons, most of that fall is in special-law offences rather than ordinary crime, while in cases the two fell alike; the count has risen in each of the last three years from a closed-border low; and the ratio to the reference population is lower than it was eleven years ago.
