Executive summary

England and Wales have repeatedly come within a few thousand places of a full prison system. Ministers have responded with accelerated release, police-cell contingencies and an urgent building programme. These measures may prevent an immediate breakdown, but they do not solve the structural problem: the number and length of custodial sentences have grown faster than the system’s safe, staffed capacity.

Norway and Japan show that a rich democracy does not have to operate its prisons at the edge of collapse. They do not, however, offer a single model. Norway confines far fewer people and deliberately moves suitable sentences into the community, including extensive use of electronic monitoring. Its correctional philosophy treats release and reintegration as the normal destination of a sentence. Japan’s prison population has fallen sharply for different reasons, including a long decline in recorded crime and prison admissions, demographic change and the retention of an estate built for a much larger population. Japan’s prisons are orderly but often highly restrictive; low occupancy should not be mistaken for proof that every feature of the Japanese regime is humane or transferable.

Oklahoma provides a different lesson. The state became an international symbol of very high incarceration, driven in part by punitive sentencing and drug policy, before voters and legislators pursued reforms. Its experience illustrates the danger of allowing imprisonment to become the default response and then relying on clemency, emergency powers or hurried releases once capacity and cost become politically intolerable. It is a warning about governing by pressure gauge.

As of 27 July 2026, England and Wales held 86,267 people in prison, or about 136 per 100,000 residents. Norway held 3,019, or 54 per 100,000, at the end of July 2026. Japan held 41,232, or 33 per 100,000, in mid-2025. These figures are not simple league tables: laws, recording practices, crime patterns and demographic structures differ. But the scale of the gap is too large to dismiss. England and Wales use imprisonment much more heavily than either comparator. (World Prison Brief: England and Wales, Norway, Japan)

Reform should not mean indiscriminate release. Serious and dangerous offenders must be confined where custody is necessary for public protection and punishment. Reform means reserving expensive prison places for those cases, reducing the use of short and ineffective custody, strengthening community sentences, making progression and release predictable, and investing in probation, treatment, housing and employment. It also means publishing a binding annual prison-demand plan so that sentencing legislation cannot quietly create more prisoners than the state can safely hold.

This report recommends that the Government:

1. establish a statutory prison-demand and capacity duty, with independent forecasts and a requirement that new sentencing laws identify their prison and probation consequences;
2. create a presumption against short custodial sentences, with tightly defined judicial exceptions;
3. expand credible, properly funded community sentences, treatment and electronic monitoring;
4. introduce earned progression for eligible determinate sentences, with exclusions and risk controls for serious violence, sexual offending and terrorism;
5. repair probation before transferring large volumes of work from prison;
6. reduce unnecessary remand and recall, while improving court speed and licence decision-making;
7. make rehabilitation operational—through purposeful activity, family contact, health care, housing and employment—not merely rhetorical;
8. retain sufficient headroom so emergency release is exceptional rather than a recurring management tool; and
9. evaluate reforms by reconviction, harm, compliance, victim confidence and total cost, not by prison numbers alone.

1. The question Britain should be asking

Public debate often begins too late. When prisons are nearly full, the apparent choice is between building cells and releasing prisoners early. Both options become politically charged: construction is presented as proof of toughness; release as proof of recklessness. Yet by the time this argument begins, the most consequential decisions have already been made—in criminal law, charging, remand, sentence lengths, release rules, recall practice and the funding of probation.

England and Wales do not have one homogeneous “UK prison system”. Criminal justice is devolved. Scotland and Northern Ireland operate separate systems, so the immediate capacity crisis and the statistics in this report concern England and Wales unless stated otherwise.

The prison population here stood at 86,267 on 27 July 2026. On a comparable whole-population measure that was roughly 136 people per 100,000, compared with 54 in Norway and 33 in Japan. The remand population in England and Wales was 15,386—17.9% of all prisoners. (World Prison Brief)

The Ministry of Justice’s last five-year projection placed the March 2029 population between 95,700 and 105,200, with a central estimate of 100,800. Projections are conditional, not destiny, but they reveal the direction produced by current policy assumptions. (Ministry of Justice prison population projections)

The House of Commons Library also records a change in the composition of the prison population: by June 2025, 48% of determinate-sentence prisoners were serving terms longer than four years, up from 36% in 2010. In September 2025, 56% of prisons were classified as crowded. (House of Commons Library) The pressure is therefore not explained simply by a sudden wave of short sentences. Longer terms, remand, recall and system delay all matter.

This is why “crime went up” is an inadequate diagnosis. Prison population is a stock. It rises when inflows grow, outflows slow, or both. Even if the number sentenced to custody were stable, longer average time served would increase the population. Conversely, crime can fall without promptly emptying prisons when long sentences already imposed continue to run.

2. Four systems, four different stories

Sources and caveat: National figures are from the World Prison Brief, which compiles mainly official data. Oklahoma includes state prisons and a complex local-jail/federal landscape, so it should not be ranked mechanically against national prison systems. Dates and definitions vary. The comparison is diagnostic, not a claim that one number captures public safety or justice.

England and Wales: a demand-management failure

The Independent Prison Capacity Review, led by Dame Anne Owers, concluded that the crisis arose from failures to align demand and supply and called for systems that trigger preventive action, including sufficient probation and community-service capacity. (Independent Prison Capacity Review) That finding matters because it reframes overcrowding. It was not an unforeseeable natural disaster. Successive policy choices accumulated without a reliable mechanism to reconcile their consequences.

Emergency release can be necessary when the alternative is refusing lawful admissions or holding people in police cells. But repeated emergency release has serious costs. Victims receive short notice. Probation teams inherit cases before plans are ready. Local authorities and charities scramble for accommodation. Prisoners may leave without treatment, documents, work or stable support. The result can be both less humane and less safe than a planned shorter sentence with properly funded supervision.

Conditions inside crowded establishments also weaken the very functions invoked to justify imprisonment. The 2025–26 annual report of HM Chief Inspector of Prisons drew on 84 inspections and related visits. Its continuing scrutiny of safety, respect, purposeful activity and rehabilitation demonstrates that nominal capacity is not enough: a place requires staff, health care, education, workshops and functioning infrastructure. (HM Chief Inspector of Prisons annual report 2025–26) A cell added to a spreadsheet is not necessarily a usable, decent or rehabilitative place.

Norway: less custody and a different use of custody

Norway is often reduced to photographs of attractive cells. That misses the structural lesson. The system uses fewer prison places because it limits their use and executes many eligible sentences under supervision outside an institution.

Statistics Norway recorded 3,359 people in the prison population on 1 January 2024. During that year, there were 2,977 new institutional prison sentences and 2,967 new prison sentences served through electronic monitoring—almost equal numbers. Electronic monitoring may be used for eligible unconditional sentences of up to six months or for the final six months before expected release, with offence-based exclusions. (Statistics Norway)

That is not “letting everyone off”. It is punishment with restrictions, surveillance and enforceable conditions, delivered without the automatic disruption of housing, employment and family ties that short imprisonment causes. Norway still uses closed prisons and preventive detention for people assessed as dangerous. Its useful principle is proportionality: the restriction should not exceed what punishment and safety require.

Norway’s low reconviction figures are frequently advertised as a single “20% recidivism rate”. Comparisons of recidivism are treacherous because countries use different follow-up periods, populations and outcomes—rearrest, reconviction or return to prison. A systematic international review found a two-year reconviction rate of 17.6% for released prisoners in Norway, but the number should not be compared casually with a differently defined English statistic. (Yukhnenko et al., 2023) The defensible lesson is not that Norwegian prisons magically produce an exact percentage. It is that normalisation, continuity of services and planned resettlement are plausible parts of a lower-custody, lower-reconviction system.

Japan: low occupancy, but not “Norway in Asia”

Japan held 41,232 prisoners in mid-2025, a rate of 33 per 100,000. Its prison population peaked at more than 81,000 in the mid-2000s and has since nearly halved. The present estate therefore contains substantial capacity designed for a larger population. (World Prison Brief: Japan)

This decline reflects a wider long-term reduction in recorded offending and in people entering prison, combined with demographic change. It cannot be attributed simply to comfortable prisons or expansive community supervision. Japanese prisons have traditionally imposed strict discipline, repetitive work and detailed behavioural rules. A reform argument should not romanticise those conditions.

Japan also faces a distinctive challenge: an ageing prison population. Older people form a growing share of admissions even as admissions among younger groups have fallen. Poverty, isolation and repeated petty theft feature in research on older prisoners. This exposes an important limit to prison policy: unused cells do not resolve the welfare failures that bring vulnerable people into custody. The lesson for Britain is to divert health and social-care needs before they become expensive penal cases, not to copy institutional rigidity.

Oklahoma: the cost of reacting after expansion

Oklahoma is relevant precisely because it is not Norway or Japan. It sits within a US system that divides people among state prisons, county jails and federal custody, uses different sentence structures, and has historically incarcerated at exceptionally high rates. Direct numerical comparison with England and Wales would create false precision.

Its policy history nevertheless makes three points. First, prison growth is strongly shaped by law: felony classifications, mandatory terms, drug sentencing, parole eligibility and revocation rules. Second, high incarceration carries fiscal and human costs without guaranteeing proportionate gains in safety. Third, once prisons and jails fill, government tends to reach for back-end relief—credits, commutations, emergency release or contracts for beds—that is less coherent than changing unnecessary admissions at the front end.

Oklahoma voters’ approval of State Question 780 in 2016, which reclassified specified low-level drug-possession and property offences, showed how a jurisdiction can reduce pressure by changing who receives a felony and prison exposure. The broader warning is applicable to England and Wales: sentencing expansion is easy to announce one offence at a time, while its cumulative prison impact remains hidden until the system reaches crisis.

3. What comparison can—and cannot—prove

Comparative penal policy is full of tempting but weak claims. A lower imprisonment rate does not by itself prove that a country is safer, fairer or better at rehabilitation. Recorded crime depends on reporting and classification. Conviction depends on policing and prosecution. Prison population depends on legal thresholds and sentence length. Occupancy depends on how capacity is defined. Reoffending figures depend on the measure and follow-up window.

Norway is smaller, wealthier per head and supported by a broad welfare state. Japan has lower levels of several forms of recorded violent crime, a shrinking population and distinctive social institutions. Oklahoma is a subnational US jurisdiction. England and Wales cannot import any system wholesale.

But comparison remains valuable when it reveals mechanisms rather than rankings. Norway demonstrates that electronic monitoring and community execution can carry a substantial share of short custodial punishment. Japan demonstrates that prison populations can fall dramatically and leave spare capacity. Oklahoma demonstrates that high incarceration is a political product that can be altered by law. Together they refute the idea that chronic near-full occupancy is an unavoidable feature of a modern democracy.

4. Why emergency release is the wrong permanent policy

The objection to emergency release is not that every recipient remains dangerous. Many prisoners would be released soon in any event. The objection is that the date is moved because the estate is full rather than because a transparent sentencing framework, risk assessment and resettlement plan say it should be.

This undermines legitimacy in four directions:

• victims can reasonably feel that the sentence announced in court did not mean what it appeared to mean;
• prison staff lose time needed for orderly progression and release planning;
• probation services receive sudden and unstable workloads; and
• the public is invited to confuse a logistical emergency with a considered reform.

Planned reform is safer than improvised release. Parliament should determine in advance what proportion of eligible sentences is custodial, what proportion can be earned through good progress, what exclusions apply, and what supervision follows. The prison service should then operate with headroom, not wait until only a few hundred places remain.

5. A reform programme for England and Wales

5.1 Put prison demand under statutory control

An independent body should publish an annual ten-year forecast covering prisons, probation, courts and community treatment. Every Bill and major sentencing guideline should include independently validated estimates of its effect on:

• prison receptions and average time served;
• remand and recall populations;
• probation caseload and required staffing;
• capital and operating costs; and
• distributional effects, including women, young adults and minority groups.

If policy creates demand above safe capacity, ministers should be required to identify funding or an offsetting reduction. This would make the hidden cumulative effect of “tougher” individual measures visible before enactment.

5.2 Presume against short custody

Short prison sentences often remove a person from work, housing, treatment and family while providing too little time for meaningful intervention. Courts should retain custody where immediate confinement is necessary, but legislation should create a presumption that sentences below a defined threshold are suspended or served in the community.

Exceptions should be explicit and reviewable—for persistent breach, significant risk, domestic abuse cases requiring protection, or other circumstances in which a community order cannot safely manage the case. A presumption is preferable to an absolute ban because it preserves judicial judgment while changing the default.

5.3 Make community punishment credible

A community sentence that begins months late or offers no treatment will not command judicial confidence. Reform therefore requires capacity before diversion. Courts need rapid access to pre-sentence information and a real menu of sanctions: unpaid work, curfews, exclusion zones, alcohol monitoring, drug and mental-health treatment, restorative processes where appropriate, and intensive supervision.

Norway’s scale of electronic monitoring is instructive, but technology is not rehabilitation. A tag can enforce time and place; it cannot secure a tenancy, treat addiction or create employment. Monitoring should be proportionate, time-limited and combined with human supervision and services.

5.4 Introduce transparent earned progression

For eligible determinate sentences, movement through custody and into supervised release should depend partly on engagement, conduct and risk reduction. Rules must be clear at sentence, not invented during a capacity crisis. Serious violent, sexual and terrorist offences require distinct safeguards and, in appropriate cases, Parole Board scrutiny.

Earned progression is defensible because it creates incentives and makes release preparation continuous. It is not defensible if “good behaviour” merely means passive compliance in a prison that offers no courses or work. The state must provide meaningful opportunities before it conditions progression on taking them.

5.5 Repair probation and reduce churn

Prison reform can fail by moving an unmanageable caseload through the gate. Probation needs stable staffing, manageable caseloads, specialist domestic-abuse and sexual-offending capacity, treatment access and timely enforcement.

Recall should protect the public when risk materially increases, but avoidable technical breach and administrative delay should not create a revolving door. Government should publish recall reasons, time spent recalled and outcomes in sufficient detail to distinguish necessary protection from system churn.

5.6 Reduce unnecessary remand

Almost 18% of prisoners in England and Wales were on remand in July 2026. Some present risks that only custody can manage. Others remain because cases are delayed or because safe accommodation and supervised bail are unavailable. Investment in court capacity, bail information, supported housing and electronic monitoring can reduce unnecessary detention without weakening public protection.

Remand reform also matters for legitimacy: people awaiting trial have not been convicted, while those awaiting sentence may ultimately receive a non-custodial disposal or a term shorter than time already served.

5.7 Turn prisons into release institutions

Most prisoners will return to the community. Public safety therefore depends on what happens before and after the gate. Each prisoner should have a coordinated plan covering identification, benefits, housing, physical and mental health, substance treatment, family contact, education and work. Progress should be measured, and agencies should share accountability for failures.

The Norwegian principle of “normality”—life inside should resemble lawful life outside as far as security permits—is valuable because release requires ordinary capabilities. Responsibility, appointments, work and constructive relationships are not indulgences. They are practice for citizenship.

5.8 Preserve genuine headroom

Official capacity is not safe operating capacity. Cells become unavailable through maintenance, violence, segregation needs and population mix. A national reserve should therefore be defined as a percentage of usable, staffed places and protected by escalation rules. Crossing thresholds should trigger published ministerial actions well before emergency conditions.

Construction remains necessary to replace dilapidated accommodation and meet irreducible demand. But new places should not become a licence for further sentence inflation. Capital plans must include staffing, health care and purposeful activity; otherwise Britain builds warehouses rather than functioning prisons.

6. Public safety, victims and the strongest objections

“Reform means dangerous people will be released”

It should not. A serious reform plan differentiates by offence, risk and sentence type. It reserves prison for people who require confinement and applies enhanced scrutiny to high-risk release. Reducing short custody for low-risk cases can protect capacity for the people who genuinely need to be held.

“Longer sentences deter crime”

The certainty of detection and proportionate punishment may deter. It does not follow that every increase in sentence length produces a matching safety gain. Marginal years are expensive, may have diminishing deterrent value and can make eventual reintegration harder. Government should be required to demonstrate the expected benefit of sentence increases, not merely their symbolic force.

“Community orders are soft”

Poorly enforced orders are weak; well-designed ones are demanding. Restrictions, frequent reporting, treatment, work and monitoring can impose real punishment while preserving the stabilising factors that reduce offending. Breach must have swift, proportionate consequences.

“Britain is not Norway or Japan”

Correct—and it does not need to be. The proposal is not cultural imitation. It is to adopt tested mechanisms, adapt them to domestic law, pilot them, publish outcomes and stop policies that fail. The relevant question is not whether Britain can become Norway; it is why Britain should keep running prisons at emergency pressure when other democracies show that different equilibria are possible.

Victims must not be an afterthought

Predictable release rules, timely notification and clear licence conditions are essential. Victims should be told what the sentence means at the outset and informed of material changes. Restorative justice should be available only when genuinely voluntary and safe. Savings from reduced custody should partly fund victim services, because penal reform loses legitimacy if it appears to transfer every benefit to the offender while externalising harm.

7. Measuring whether reform works

Prison population reduction is a means, not the ultimate outcome. A public dashboard should track:

• proven reoffending at consistent one- and two-year intervals;
• frequency and seriousness of reoffending, not just whether any offence occurred;
• compliance and completion for each community intervention;
• victim satisfaction, notification and protection outcomes;
• deaths, self-harm, violence and use of force in custody;
• hours out of cell, education, treatment and employment;
• housing and employment after release;
• racial, sex and age disparities;
• prison and probation staffing; and
• full fiscal cost across justice, health, housing and benefits.

Reforms should be phased and evaluated with credible comparison groups where possible. Results should be published even when disappointing. Norway’s or Japan’s headline rate cannot substitute for domestic evidence.

Conclusion: govern the pipeline, not the panic

England and Wales are not short of emergency ingenuity. They are short of a durable rule connecting sentencing policy to safe capacity. When that connection breaks, government alternates between expensive construction and hurried release, while staff, victims, prisoners and communities absorb the consequences.

Norway shows the value of limiting institutional custody and making supervised alternatives ordinary. Japan shows that low occupancy can follow a sustained fall in admissions and that demographic and social policy matter to penal outcomes. Oklahoma shows how punitive laws can build a high-incarceration system—and how reform must address the front door rather than depend on relief at the back.

The responsible reform position is neither “release everyone” nor “build forever”. It is to imprison fewer people unnecessarily, confine dangerous people safely, make every sentence intelligible, invest in supervision and reintegration, and operate with enough headroom that release dates are governed by law and risk rather than the number of empty cells on a Tuesday night.

That is not softness. It is the harder discipline of making punishment effective, proportionate and governable.