Does Prison Work?

There are three different claims being made when we talk about the effectiveness of prison and it is important to distinguish them:

• that the threat of prison discourages people from committing offences in the first place;

• that actually imprisoning an offender discourages that individual from offending again; and

• that imprisonment changes an offender in such a way that he or she is less likely to reoffend after release.

The evidence for these propositions is quite different.

As a means of incapacitation, prison unquestionably works. A person who is locked up has considerably less opportunity to commit offences against the general public. This is particularly important with persistent and dangerous offenders. But incapacitation is not the same thing as deterrence. It tells us nothing about what the prisoner would have done had he or she remained at liberty, nor does it tell us whether the person will offend after release.

The evidence for general deterrence---the idea that the threat of imprisonment causes people in general to refrain from crime---is surprisingly uncertain. It is reasonable to suppose that punishment has some deterrent effect: people normally take account of the possibility of being caught and punished. But increasing the severity of imprisonment does not appear to produce a correspondingly large reduction in crime. Research in criminology has generally found that the certainty of being caught is a more powerful deterrent than the severity of the eventual punishment. A person contemplating a burglary is unlikely to calculate carefully whether the sentence will be six months, two years or five years if he thinks there is only a small chance of being caught. The perceived probability of detection may therefore matter more than the severity of the prison sentence.

There is also an important distinction between the certainty and severity of punishment. Governments have often responded to public concern about crime by increasing maximum sentences, but there is relatively little reason to believe that simply making sentences longer produces proportionately greater deterrence. Indeed, there is a theoretical paradox here: if very long sentences are imposed for relatively common offences, the additional deterrent effect may be quite small while the financial and social costs of imprisonment become very large.

The evidence concerning individual deterrence is also less impressive than one might expect. If prison were strongly deterrent to individual offenders, we would expect people who had been imprisoned to be substantially less likely to offend subsequently. In England and Wales, however, reoffending remains substantial. The Ministry of Justice's latest published statistics show that:

• 37.5 per cent of adults released from custody or starting a court order in the April–June 2024 cohort committed a proven reoffence within the relevant follow-up period.

• For those released after sentences of less than twelve months, the figure was 67.1 per cent.

• For those released after sentences of twelve months or more, it was 23.2 per cent.

These figures need to be interpreted cautiously because prisoners are not randomly selected for imprisonment. People receiving short custodial sentences tend, on average, to have more extensive criminal histories and more serious social and personal problems than people receiving community penalties. The Ministry of Justice itself warns that simple comparisons between different sentences cannot establish which sentence is more effective because the characteristics of the offenders receiving them differ. Nevertheless, a reoffending rate of 67.1 per cent after short custodial sentences certainly does not suggest that a short spell in prison is a particularly powerful individual deterrent.

Indeed, there is evidence for what might be called a criminogenic effect of imprisonment: under some circumstances imprisonment can increase rather than decrease subsequent offending. This is not especially mysterious. A prisoner can lose employment, accommodation and family relationships; acquire contacts with other offenders; become institutionalised; and emerge with fewer legitimate opportunities than before. These effects are particularly important with short sentences, where there may be little time for constructive rehabilitation but enough time for the disruptive consequences of imprisonment to take effect.

This brings us to rehabilitation, where the answer is more encouraging. The evidence does not support the proposition that prison inevitably reforms people, but there is substantial evidence that particular interventions can reduce reoffending. The Ministry of Justice's 2025 synthesis of the evidence examined interventions involving accommodation, education, employment, finance and debt, community ties and other factors associated with desistance. It specifically drew on systematic evidence reviews and meta-analyses where available.

The crucial implication is that it is often misleading to ask whether 'prison works'. Prison is an environment in which certain things can be done. Some of those things appear to work better than others. Education, vocational training, treatment for substance misuse, psychological interventions, help with accommodation and employment, and maintaining constructive family and community relationships can all contribute to reducing the probability of reoffending. But that does not establish that the deprivation of liberty itself has a rehabilitative effect.

There is particularly interesting evidence concerning community-based rehabilitation. A 2024 Ministry of Justice evaluation examined community sentences containing accredited treatment requirements for alcohol misuse, drug rehabilitation or mental-health treatment, comparing them with community sentences without such requirements and with short custodial sentences.

This is important because it illustrates a broader finding in the evidence: sometimes the most effective response to offending is not to impose more punishment but to address the particular problems that are producing the offending.

The figures on short prison sentences are therefore especially relevant to the present British debate about early release. If someone receives a six-month sentence, it is tempting to assume that six months of imprisonment must be more effective than six months of a demanding community sentence simply because prison is more severe. The evidence does not justify that assumption. A well-designed community sentence can leave the offender subject to substantial restrictions while allowing him or her to retain employment, accommodation and family relationships and, crucially, to receive treatment for the problems contributing to offending.

There is nevertheless a danger in drawing the opposite conclusion and saying that prison does not work. That would be equally unjustified. Prison clearly works through incapacitation, and for some offenders a substantial period in custody may be necessary to protect the public. Moreover, prison-based rehabilitation can work when it is properly designed and delivered. The question is not whether prison is inherently good or bad, but what we are trying to achieve by imprisoning a particular person and whether imprisonment is the most effective means of achieving it.

There is also an important moral dimension. Even if imprisonment had little rehabilitative or deterrent value, one might still maintain that people who commit serious offences deserve to be deprived of their liberty. That is the retributive justification for punishment, and it is logically independent of whether prison reduces future crime. Conversely, if the primary purpose of imprisonment is supposed to be crime reduction, then evidence of its effects on subsequent offending becomes crucial. We should not assume that the harshest punishment is necessarily the most effective punishment.

The most defensible conclusion from the evidence is therefore rather nuanced.

• Prison works extremely well as incapacitation;
• it probably has some deterrent effect, but the additional deterrent effect of making imprisonment longer or harsher appears limited;
• imprisonment by itself is not a reliable method of rehabilitation.

Rehabilitation is most successful when imprisonment is combined with interventions directed at the causes of offending and with effective support after release. Short periods of imprisonment are particularly questionable because they can disrupt the very employment, housing and social relationships that help people to stop offending.

The current British debate about releasing prisoners early is consequently not simply a choice between being 'soft' or 'tough' on crime. It raises the more fundamental question of whether the punishment that feels most severe is necessarily the punishment that does the most to prevent the next crime. The available evidence suggests that the answer is often no.