How Many Elderly People Are in Japanese Prisons?

Quick answer

At the end of 2024 there were 5,085 sentenced prisoners aged 65 and over held in Japanese penal institutions, out of a total sentenced population of 33,745. That is 15.1 per cent. Within that figure, 1,674 were aged 65 to 69, 1,516 were aged 70 to 74, and 1,895 were aged 75 or over. Six hundred and thirty-two of the 5,085 were women.

The same tables show that this number is currently falling. At the end of 2020 the figure was 5,624, so the elderly prison population is down 9.6 per cent in four years, at a time when the population of Japan aged 65 and over was still growing.

A different official figure, 2,049, is also correct and is quoted more often. It counts the people aged 65 and over who were newly admitted to prison during 2024. The two numbers are not rival estimates of the same thing. One counts everyone inside on a particular day; the other counts arrivals over a year. A person serving a six-year sentence is counted once in the admissions figure and appears in six successive year-end counts.

The direction of travel depends on which measure is used, and the measures disagree.

• Counted as a headcount, prisoners aged 70 and over rose from 1,386 at the end of 2004 to 3,411 at the end of 2024, a rise of 2.46 times.
• Counted as a share of the prison population, they rose from 2.2 per cent to 10.1 per cent, a rise of 4.67 times, though the total sentenced population nearly halved over those twenty years, from 64,047 to 33,745, and a shrinking denominator does most of that work.
• Counted against the population of the same age, the rate was 7.91 per 100,000 in 2004, 12.68 in 2014 and 11.77 in 2024. Of those three points the highest is the middle one, so on that measure the level today is below where it stood ten years ago. The tables used here do not give every intervening year, so the exact turning point is not established.

All three statements are true at once. Which one is quoted determines whether Japan appears to have a growing problem or a receding one.

On the question of why, the record is thinner than the coverage suggests. None of the official studies examined for this article asked offenders whether they committed a crime in order to be imprisoned. Four of them asked why people offended, using fixed lists of between 11 and 18 options; no such option appears in any of those lists. The one study that coded circumstances for elderly repeat property offenders recorded, most often, neither hunger nor destitution but estrangement: no family in 20 of its 33 cases. In the largest survey of women prisoners, 58.8 per cent said they had nobody to consult.

Two questions, not one

Japan publishes two counts of elderly prisoners, in two different places, and they are built to answer different questions.

The Correctional Statistics Survey counts everyone held on 31 December. This is the stock: the population of the prison on a given day. Its 2024 table gives 5,085 sentenced prisoners aged 65 and over.

The White Paper on Crime reports newly admitted prisoners: everyone who entered during the calendar year. This is the flow. Its 2024 figure is 2,049.

Neither is a better number. They measure different things, and the gap between them is what one would expect when people stay longer than a year. It cannot be turned into an average length of stay: the year-end count also includes people who were admitted below 65 and reached that age inside, so the two figures do not describe the same set of people. What the gap is not is evidence that anything is being undercounted.

The distinction matters here more than it usually would, because a third population is larger again by an order of magnitude. In 2024 Japanese police cleared 41,070 people aged 65 and over as suspects in Penal Code offences. The Japanese term, kenkyo, covers cases sent to prosecutors without the suspect being taken into custody, so “arrests” is a loose translation of it.

Those 41,070 are not the pool from which the year’s 2,049 new prisoners were drawn. They are a separate count, in a separate system, with its own timing: a case cleared in one year may produce a sentence in the next, or no sentence at all, and the prison figures cover offences under other statutes as well. The two cannot be divided into a disposal rate, and this article does not attempt one. What they show is a difference of scale: clearances involving elderly suspects run into the tens of thousands each year, while admissions to prison of people that age run to about two thousand. Any sentence that begins “elderly offenders in Japan” needs to say which of the three groups it means.

The same problem of unlabelled denominators runs through the share of bodies Japan autopsies and the count of people who go missing: a figure attached to one base, repeated until the base falls away.

Twenty years, three answers

Here are the three measures side by side, for prisoners aged 70 and over. That age band is used because it is the only one that can be tracked consistently: the current three-way split of 65–69, 70–74 and 75-plus appears in the tables only from 2020, and the earlier tables offer just 60–69 and 70-plus.

Three things happened at once, and they pull in different directions.

The prison population collapsed. Counting all categories of detainee, Japan held 81,255 people at the end of 2006 and 40,544 at the end of 2024. The sentenced population, which is the series used throughout this article, fell from 64,047 at the end of 2004 to 33,745 at the end of 2024. This is the largest movement in the table and it has nothing to do with age.

The elderly population grew. There were 17.5 million people aged 70 and over in Japan in 2004 and 29.0 million in 2024, a rise of 1.65 times. A country with 65 per cent more people in an age band will, all else equal, have more prisoners from that band.

Elderly prisoners rose by more than population growth alone explains, and then the excess stopped growing. The 2.46-fold rise in headcount, set against the 1.65-fold rise in population, leaves a 1.49-fold rise in the rate over twenty years. That residual rise did not continue: between 2014 and 2024 the rate per 100,000 people aged 70 and over fell by about 7 per cent, from 12.68 to 11.77.

The 4.67-fold rise in the share of the prison population, which is the figure most likely to be quoted because it is the most dramatic, is the product of a numerator that grew 2.46 times and a denominator that shrank to 53 per cent of its former size. It is a true statement about the composition of Japanese prisons. It is not a statement about elderly offending, and it should not be read as one.

Clearances point the same way as the rate. People aged 65 and over cleared for Penal Code offences peaked in 2008 at 48,786 and stood at 41,070 in 2024. For women aged 65 and over the peak was 2012, at 16,502, against 12,625 in 2024, a fall of more than a fifth. Newly admitted elderly prisoners peaked in 2016 at 2,498, against 2,049 in 2024. Each of the three flow measures cited here peaked before 2017.

Why the prison population halved

The denominator in that table is doing so much work that it deserves its own paragraph, because a reader who does not know why Japanese prisons emptied will misread everything that follows.

Recorded crime in Japan fell steeply over two decades. Police recorded 2,854,000 Penal Code offences in 2002. By 2021 the figure had fallen to about 568,000, the lowest of the postwar period, and although it has risen for four consecutive years since, the 2025 total of 774,142 is about 27 per cent of the 2002 peak. Japan is coming off the floor rather than returning to a dangerous past, and the prison population reflects the two decades of decline that preceded the floor.

Fewer offences mean fewer prosecutions, fewer custodial sentences and, with a lag of several years as sentences run their course, a smaller prison population. Decisions taken after the offence work in the same direction: a large share of cases is disposed of by suspended prosecution, which is how two thirds of cleared suspects avoid a charge, and this article does not attempt to apportion the fall between offending and disposal. The sentenced population fell from 64,047 at the end of 2004 to 33,745 at the end of 2024.

The size of that fall is why the share of elderly prisoners rose so steeply, and it is why the share should not be used as a measure of elderly offending. If a prison holds a hundred people, ten of them old, and eighty of the young ones leave and are not replaced, the old share of the prison has risen from 10 per cent to 33 per cent without a single additional old person being admitted. Japan did not experience that in its pure form, since the number of elderly prisoners did rise, but a little under half of the movement in the share is arithmetic of exactly this kind: of the 4.67-fold rise, the shrinking denominator accounts for about 42 per cent and the growing numerator for about 58 per cent.

Inside “elderly”, the directions split

The category “elderly” hides its own reversal. Comparing the end of 2020 with the end of 2024, the only years for which the three-way age split exists, the bands move in opposite directions.

Only the oldest band is growing. The two younger elderly bands are falling faster than the prison population as a whole, and the total for everyone aged 65 and over is down 9.6 per cent across the four years. And the crossover has already happened: at the end of 2024 there were more prisoners aged 75 and over (1,895) than aged 65 to 69 (1,674).

This is worth stating plainly because the aggregate figure conceals it. “The number of elderly prisoners is rising” is not true of people in their late sixties in Japan, whose numbers are down a fifth in four years. It is true of people in their late seventies, eighties and beyond.

The account that circulates in English

The English-language account of this subject is largely a body of reported feature journalism, and it is not fabricated. Reporters went to Japanese prisons, obtained access, and interviewed prisoners who told them why they were there. The BBC has reported on pensioners who said they wanted to be in jail. CNN has reported on elderly women describing loneliness and want. El País has described prisons functioning as retirement homes for low-income older people. Those interviews record real statements by real people.

What the coverage does not do is establish frequency. An interview establishes that a motive exists. It cannot establish how common it is, and the pieces do not claim otherwise; the claim tends to arrive later, in aggregation, when the individual account is restated as a general explanation of a national statistic.

That is the gap this article is about. The English-language record contains a well-evidenced set of individual accounts and an unevidenced general claim. The Japanese statistical record contains the opposite: a large body of aggregate data, and almost nothing on this particular motive, because nobody has asked.

The questions on the form

Japan has run four substantial studies that recorded, for convicted or sentenced offenders, the reasons behind the offence. Their categories can largely be listed, and listing them is the point.

Research Department Report 37 (2007) is the only one that asked prisoners directly. Questionnaires went to 607 elderly prisoners approaching release; 391 replied, a response rate of 64.4 per cent. Question 10 asked why they had come to be in prison, with multiple answers permitted from 18 options:

being careless with money · not being able to stop drinking · life being hard · having no work · being deceived by someone · vanity · acting out of desperation · idleness or gambling · living beyond one’s means · having no trade · bad company · starting to use stimulants · being abandoned by family · a failed relationship · becoming a yakuza · a bad neighbourhood · bad parents or family · other.

Research Department Report 56 (2017) tabulated circumstances for 33 elderly repeat property offenders, from records rather than interviews, using 11 categories: estrangement from family or absence of relatives · habitual stealing · financial hardship · acquisition for personal use · thrift · poor health · habitual drinking or alcohol dependence · hunger · idleness · illness or death of a relative · gambling.

Research Department Report 57 (2017) coded motives and background circumstances for 1,376 shoplifting offenders from case records, with housing instability among the background options.

The 2018 White Paper study of 2,421 people convicted of theft used nine motive options and nine background options, the latter including housing instability and estrangement from family.

In three of the four studies, and among the categories published for Report 57, we found no option corresponding to wanting to enter prison, wanting to return to prison, having nowhere to live, or having nothing to eat as an end in itself. The nearest available categories are hunger, in two of the studies, and housing instability, as a background circumstance rather than a motive. Report 57 publishes only its top five categories, so its full coding frame is not on the record.

The one questionnaire that prisoners filled in themselves, Report 37, ended its list with “other”, so a man who wanted to say it was not silenced. But an “other” that is never broken out in the published tables cannot be counted, and the report does not publish what went into it. The remaining studies did not ask anyone anything: they sorted existing records into residual categories whose contents were also not published. A motive that has no box of its own, offered on a form filled in inside the institution the respondent would be saying he wanted to be in, is not a motive these instruments were built to find.

Two individual descriptions of the familiar kind do appear in official documents. The 2008 White Paper carries a case summary of a man aged 76 who said that prison offered a bed, meals and work, and who wanted to be arrested. Report 56 carries one case of a prisoner who felt constrained in the community and reoffended in order to return to prison. That man was in the chapter on prisoners with mental disorders, not the chapter on elderly prisoners. Neither case is part of any tabulation.

So the accurate statement is narrow, and it is worth being precise about what it does and does not mean. None of these studies measured the frequency of this motive, and we have found no Japanese government survey that has. That is a fact about what the state chose to ask, and about how far we were able to search. It is not evidence that the motive is rare, and it is certainly not evidence that the prisoners interviewed by foreign correspondents were misrepresenting themselves.

Why the question was never on the form

The omission is not obviously deliberate, and the structure of the studies suggests a duller explanation.

Three of the four exercises are not surveys of people at all. Reports 56 and 57 and the 2018 White Paper study code information out of case records: prosecution files, judgments, prison classification documents. A case file records what an offender said about the offence at the time, filtered through what an investigator thought worth writing down. A defendant explaining a shoplifting charge to a prosecutor has strong reasons to describe hardship and weak reasons to volunteer that he was aiming for a custodial sentence, and an investigator has no box in which to record it if he did. The categories in those studies are the categories the files already contain.

Only Report 37 asked prisoners directly, and its 18 options read as a list of personal failings and misfortunes (drink, vanity, idleness, bad company, being deceived), which is a framing built to explain why someone fell into crime, not to capture a deliberate choice of prison as a destination. The option was missing because the question was conceived as being about causes of failure rather than about calculated preferences.

This is inference from the design of the forms, not from any statement by the people who designed them. What is not in dispute is the consequence: the one claim that dominates the English-language coverage of this subject is the one claim on which Japan has published no frequency data at all.

What the studies did record

Set the missing option aside and read what the four studies did record. The results are consistent with each other and they are not the results the English-language account would predict.

In Report 37, the 391 prisoners who answered chose, most often:

Hardship is present and it is third. It is also a minority answer: three-quarters of these prisoners did not select it. Among the 182 in the sample imprisoned for theft it rises to first place, but it does not become a majority.

In Report 56, among the 33 elderly repeat property offenders, the most frequently recorded circumstance was not financial at all:

Thirty-three is a small number and it is a record-based coding rather than a survey of the men themselves, so it carries less weight than Report 37. But it points where the larger study points.

Report 57 is much larger and points the same way on hardship. Among 174 men aged 65 and over convicted of shoplifting, the coded motives were acquisition for personal use in 60.9 per cent of cases, thrift in 55.2 per cent, financial hardship in 27.0 per cent, habitual stealing in 19.5 per cent and hunger in 17.2 per cent. Thrift outranks hardship, and hunger appears in about one case in six. In a subset of 71 men aged 65 and over dealt with by a fine, hardship was coded in 17 cases and hunger in 10; of those 10, four were reconvicted of theft.

The 2018 White Paper study, covering 2,421 people convicted of theft in a single month of 2011, offered nine motive options and nine background options. Financial hardship was recorded for about one elderly man in four. Its background list is the one that contains housing instability and estrangement from family, and those two categories recur across every study here that offered them.

The clearest evidence comes from the Ministry of Justice special study of women prisoners, published in the 2023 White Paper and available in the ministry’s own English edition. Theft accounted for 57.7 per cent of the women prisoners in that study, and among those aged 60 and over, 95.9 per cent of the theft cases were shoplifting. That is the group at the centre of the international coverage. Their recorded circumstances were:

These three come from separate items in the study rather than from a single question with one set of alternatives, so they should be read as three independent proportions and not as shares of one whole. What can be said is that the item about having no one to consult drew the highest positive response of the three.

The 37.4 per cent is lower than the corresponding figure for women prisoners under 60 and lower than the figure for men. On the item that drew the highest response, well over half of these women recorded having no one to turn to.

The oldest of the studies points the same way on the question of sex differences, though with numbers small enough that it can only be suggestive. In a 2007 review of 139 elderly theft offenders handled by the Tokyo District Public Prosecutors Office, financial hardship was recorded for 74 of the 112 men (66.1 per cent) but for only 6 of the 27 women (22.2 per cent); for the women the most frequent entries were wanting the item itself, in 17 of 27 cases, and thrift, in 16. That is one prosecutor’s office, in one year, with 27 women in it. The 2023 national study, which is larger and current, produced a higher figure for hardship among older women, 37.4 per cent, and the two should be read together rather than either being taken alone.

Read across the studies, the picture that emerges is not quite the one in circulation. Material need appears in roughly a quarter to two-thirds of cases, depending on the study, the sex of the offender and the wording: 24.6 per cent of the prisoners in Report 37, 16 of the 33 cases in Report 56, 27.0 per cent of men aged 65 and over in Report 57, 37.4 per cent of women aged 60 and over in the 2023 study, and 66.1 per cent of the men in the 2007 Tokyo review. In the two studies that coded the absence of family or of anyone to consult, it appears in a majority: 20 of 33 cases, and 58.8 per cent.

The English-language account pairs loneliness and poverty as a single explanation. In the Japanese data the two come apart, and they do not fare equally. The isolation half is supported wherever it was measured. The poverty half is present in every study and dominant in none of the ones that asked older women.

None of this means the hardship is imaginary. A study commissioned by the Tokyo Metropolitan Government and reported in 2017 compared 56 elderly shoplifting suspects who had been dealt with by the lightest available disposal against 1,336 members of the general elderly population. Among the suspects, 44.6 per cent said they felt their present life was hard, against 17.7 per cent of the comparison group: a real and large difference in how people describe their own circumstances, even though the samples are small. The same survey recorded that 87.5 per cent were not receiving public assistance and 50 per cent owned their homes. Those two facts sit awkwardly together, and it would be wrong to resolve the tension by declaring the hardship unreal: take-up of public assistance in Japan is known to be far below eligibility, and a suspect dealt with by the lightest available disposal is not a representative offender. What can be said is that the subjective and objective indicators diverge, which is a different and more interesting finding than either “they are poor” or “they are not”.

The women

The international coverage of this subject is disproportionately about women, and on the numbers that emphasis is defensible even though the absolute figures are small.

Of the 5,085 sentenced prisoners aged 65 and over held at the end of 2024, 632 were women. Women are a small minority of the Japanese prison population overall, about 2,900 of 33,745, so the elderly share among women prisoners is much higher than among men. Among newly admitted women prisoners in 2024, 21.6 per cent were aged 65 or over, against 13.8 per cent across all admissions. Twenty years ago the figure for women was 6.2 per cent, a rise of 15.4 percentage points.

The growth is concentrated at the oldest ages. Newly admitted women prisoners aged 70 and over numbered 55 in 2005 and 253 in 2024, a rise of 4.6 times, the steepest movement anywhere in these tables. At year end, women aged 75 and over in prison numbered 309 of the 632.

What they are in for is almost uniform. Among women prisoners aged 60 and over in the ministry’s study, theft was the offence in the majority of cases and 95.9 per cent of those thefts were shoplifting. Among elderly women cleared by the police rather than imprisoned, around nine in ten are cleared for theft, and around eight in ten of those thefts are shoplifting.

So the figure that appears in the coverage, an elderly woman imprisoned for taking goods from a shop, is a real and dominant type, not a colourful exception. The part that does not survive contact with the data is the explanation attached to it. In the study, women aged 60 and over reported problems with living expenses less often than women under 60 and less often than men, at 37.4 per cent, and reported having nobody to consult at 58.8 per cent.

There is also a caution about numbers this small. Two hundred and fifty-three admissions in a country of 124 million people is not a social phenomenon of the scale the coverage implies, and a 4.6-fold rise from a base of 55 is a large multiple of a very small number.

What the prison does with them

Japanese prisons were not built for this population and have been visibly adjusting to it.

In 2024 the correctional system administered dementia screening to 2,086 prisoners, referred 284 of them for examination by a doctor, and recorded a diagnosis of dementia in 132 cases. The screening programme exists because the need was identified; the diagnosed number is also small, and nothing in these figures supports treating elderly prisoners in general as a cognitively impaired population. The 2,086 screened are not stated to be all elderly prisoners, so 132 should not be divided by 5,085.

The Correction Bureau has also introduced a treatment course for elderly prisoners that covers welfare arrangements for life after release, alongside the existing arrangements under which prisoners with no home to go to can be referred to welfare authorities before they leave.

That last provision is the most telling thing in the corrections chapter, because of what it implies about who is arriving. A prison system that has to arrange housing and welfare registration for its departing prisoners is not processing people who have somewhere to return to. It is the administrative shadow of the same finding that runs through the motive studies: 20 of 33 elderly repeat property offenders estranged from family or without relatives, 58.8 per cent of older women prisoners with nobody to consult.

Where the familiar account is right

One part of the story is not a myth, and the statistics support it more strongly than they support anything else in this article.

Elderly prisoners in Japan return. The 2007 survey recorded an average of 6.4 prison admissions per prisoner among those it studied, 6.8 for men and 2.3 for women, with one man recorded at 41 admissions and one woman at 11.

The same report followed release cohorts from 1996 to 2005. Among prisoners released at the expiry of sentence rather than on parole, those aged 65 and over were readmitted within five years at an estimated 62 to 70 per cent, on a denominator adjusted for estimated deaths, and the rate was higher than for those under 65. Those are historical cohorts, and the figure does not describe the people admitted today.

Whatever is driving elderly imprisonment in Japan, the prison population it produces is not made up of one-off offenders. It cycles. This is a statement about the people who reach a prison, not about the tens of thousands cleared as suspects each year, and the admission counts behind it were measured in the 2000s. The three most frequently recorded circumstances in Report 56, estrangement from family in 20 cases of 33, habitual stealing in 18 and financial hardship in 16, describe someone with no one outside, an established pattern, and not enough money, in that order.

There is a second point on which the familiar account holds, and it concerns who gets out early. The 70 per cent figure above applies to prisoners released at the expiry of their sentence rather than on parole; among those released on parole, the reimprisonment rate showed no comparable gap by age.

Two readings of that gap are available and the published tables do not choose between them. One is selection: parole boards assess risk, and if they release the prisoners least likely to reoffend, a gap of this kind will appear whatever the role of family. The other is the one that fits the rest of this article: parole in Japan is easier to obtain with somewhere to go and someone to receive you, so an elderly prisoner with no family is more likely to serve to the end, leave without a release plan, and appear again in a later year’s admissions.

If the second reading is right, it is the same isolation showing up a third time, now as a feature of the release system rather than of the offence. That is a reading of two statistics placed side by side, not a documented causal finding, and we cannot separate it from the first.

So how many?

For anyone quoting a figure, these are the numbers with their definitions attached.

If the question is how many elderly people are in prison in Japan today, the answer is 5,085 sentenced prisoners, a figure that excludes people held before trial, and it is falling in the two younger elderly bands while rising among those aged 75 and over. If the question is whether elderly imprisonment is still growing relative to the number of elderly people in the country, the answer is that the rate is lower now than it was at the middle of the three points measured, ten years ago. If the question is why they are there, the honest answer is that the state has published a great deal about what they did and comparatively little about why, and that the single explanation most often given in English does not appear as a question on any of the forms we were able to examine.

What we could not confirm

• Whether the 5,085 figure excludes every other category of detainee. The table is headed as covering sentenced prisoners held at year end, and the total of 33,745 matches the White Paper’s sentenced population. We have not independently verified the treatment of people held in default of a fine.
• The composition of the 65-and-over group before 2020. The three-way age split does not exist in the earlier tables, so the twenty-year series in this article is for people aged 70 and over only. A 65-and-over series of the same length cannot be built from the published tables.
• The basis of the dementia screening population. The 2,086 people screened are not stated to be all elderly prisoners, and the 132 diagnoses should not be divided by the elderly prison population.
• Whether any survey outside the four described here has asked about the motive. We read Reports 37, 56 and 57 and the relevant white paper studies. Report 57 publishes only its top five categories, so a complete list of its coded options was not available to us, and the same is true of the residual categories in the record-based studies.
• How far the reimprisonment figures carry to the present. The 62 to 70 per cent applies to prisoners released between 1996 and 2005 and rests on a denominator adjusted for estimated deaths. We did not find an equivalent figure for recent release cohorts.
• Exact publication dates and wording of the English-language features discussed above are summarised from their headlines and opening paragraphs rather than quoted in full.