Inside England and Wales’s prisons crisis: Which prisons do well?

Some prisons perform much better than others – particularly open prisons

Performance is poor across many prisons, but this is not a uniform picture. There is dramatic variation between the best- and worst-performing prisons: the prisoner-on-prisoner assault rate for 2023/24 ranged from 8 per 1,000 prisoners to 914. ⁵¹ Ministry of Justice, ‘Annual Prison Performance Ratings 2023-24 Supplementary Tables’, 2024, www.gov.uk/government/statistics/prison-performance-ratings-2023-to-2024 We see similar results for other performance measures, from housing on release to protesting behaviours to participation in work or education.*

Some types of prison perform better than others, but open prisons are the clear stand-out across all metrics. Most category C training prisons have lower levels of violence and protesting behaviour and higher purposeful activity rates compared to other categories, but there are a few clear outliers. Both of these findings may be significantly driven by the different types of prisoners housed in these types of prison compared to others, as discussed below. Some other prison types have a more mixed picture. High-security prisons have relatively low levels of prisoner-on-prisoner assaults and incidents at height, but relatively higher staff assaults. Women’s prisons show a similar pattern, though with far higher staff assaults and drastically higher self-harm rates. The median staff assault rate in women’s prisons is 278 per 1,000 prisoners, by far the highest of any prison category. ⁵² Ministry of Justice, ‘Annual Prison Performance Ratings 2023-24 Supplementary Tables’, 2024, www.gov.uk/government/statistics/prison-performance-ratings-2023-to-2024

Reception prisons is consistently the poorest performing category, especially on violence and participation in purposeful activity. This group has experienced the most acute pressure and crowding during the capacity crisis and usually have more limited purposeful activity as standard, as they are intended for shorter stays. They also hold remand prisoners who are yet to be convicted and so are exempt from prison rules requiring prisoners to work (though they may choose to work if they wish).

Reception prisons are also disproportionately Victorian-era inner-city prisons, and are often in very poor physical condition: Pentonville, for example, had such a severe rat infestation in its kitchen that catering was temporarily shut down in April 2024, after the Independent Monitoring Board referred the situation to the local council environmental health service and the prisons minister. ⁵³ Independent Monitoring Boards, Breaking point: the impact of a crumbling prison estate on prisoners, 27 November 2024, retrieved 23 December 2024, https://imb.org.uk/news/breaking-point-the-impact-of-a-crumbling-prison-estate-on-prisoners  Five of the eight prisons that have received an ‘urgent notification’ in the last two years – in which the prisons inspectorate raises concerns directly to the secretary of state – have been reception prisons.

There is broad consistency in which prisons do well and badly both over time and across the range of performance metrics we look at.** Twelve prisons (10% of those for which we have data) are in the worst 20% across at least half of the metrics we look at – and seven of these are reception prisons (compared to 25% of all prisons).

We see an even stronger pattern in reverse among the best performers. Twenty-one prisons (18% of the total) are in the top 20% of prisons across at least half of the available performance metrics. This includes all open prisons, which do well across the board, but also a few from other categories, including Stafford (category C trainer), Warren Hill (category C trainer/resettlement) and Grendon (category B). The closed prisons in this group are not typical category B or C prisons: all have a particular cohort (for example, sex offenders) and/or rehabilitative purpose (such as intensive psychotherapy) that they serve. There are no reception prisons in this group. The two women’s prisons among the top performers, East Sutton Park and Askham Grange, are both open prisons.

_{* We only have data on the proportion of prisoners taking part in at least some purposeful activity, not the total amount of purposeful activity being delivered.}

_{** These metrics are: the proportion of prisoners in crowded accommodation; the prisoner-on-prisoner assault rate; staff assault rate; self-harm rate; incidents at height rate; barricade rate (when prisoners obstruct staff access to part of the prison); educational progress in English and maths; attendance at education courses; achievement of vocational qualifications; percentage of prisoners in purposeful activity; percentage employed at six weeks following custodial release; percentage employed at six months following custodial release; percentage housed on the first night after release from custody; the staff sickness rate and the staff resignation rate. See the Annex for the full table.}

Excepting open prisons, there is generally more variation within prison categories than between them

Despite these general trends, on the whole variation is greater within prison categories than between them, with the clear exception of open prisons. Some other categories are outliers on specific metrics, such as female prisons having much higher rates of self-harm and, to a lesser extent, high-security prisons having lower levels of prisoner-on- prisoner assault. Nonetheless, the interquartile ranges, which represent the middle 50% of prisons within each category, for most prison categories overlap substantially. In practical terms, this means that if you were given all the performance data for any individual prison, it would be extremely hard to identify which category it was in – unless it was an open prison.

This is a potentially surprising result. Staff levels vary substantially across prison categories, and they are often dealing with quite different prisoner cohorts. Category C training prisons, which are specifically intended to provide education and training, do mostly have higher levels of purposeful activity than most other closed (non-open) prisons. But this is still some way off the level in open prisons, and other category C prisons are not meaningfully different to other prison categories. Interviewees suggested that recent years have seen a compressing of the differences between categories. Capacity pressures mean prisoners have been placed wherever there is a cell available, rather than in the category most suitable for them, ⁵⁴ Independent Monitoring Boards, National Annual Report 2023, May 2024, retrieved 23 December 2024, https://cloud-platform-e218f50a4812967ba1215eaecede923f.s3.amazonaws.com/uploads/sites/13/2024/05/IMB-2023-National-Annual-Report-.pdf and the widening gap between good and poor performers within categories may also reduce differences between categories.

“One thing we have definitely seen is an erosion of differences between types of prisons. You used to be able to differentiate [between categories] but that’s just not true now” – voluntary sector organisation

Some of the variation in performance is driven by differences in prisoner populations… 

Prison performance is substantially affected by the specific cohort of prisoners in each prison. Those serving long sentences, particularly those with no fixed release date such as life sentences, are often considered to be more stable and less likely to become involved in violence. ⁵⁵ McGuire J, Understanding prison violence: a rapid evidence assessment, HM Prison and Probation Service, 2018, https://assets.publishing.service.gov.uk/media/5b8e434540f0b67d970b92d5/understanding-prison-violence.pdf ⁵⁶ Institute for Government interviews, 2025. Grendon, for example, is the best-performing category B prison across a host of metrics, and also has a lower reoffending rate than similar prisons. ⁵⁷ HM Chief Inspector of Prisons, Report on an unannounced inspection of HMP Grendon, 2023, www.justiceinspectorates.gov.uk/hmiprisons/wp-content/uploads/sites/4/2023/08/Grendon-web-2023.pdf  But it has a highly selective offender cohort. It is a dedicated therapeutic prison, and prisoners have to volunteer to transfer to Grendon. Those who are considered ‘not ready’ for therapy or who do not engage with their therapeutic programme may be transferred out, and most offenders are serving long sentences. ⁵⁸ Williams Z, ‘Life inside a therapeutic prison: ‘Look, we’ve done some terrible things…’’, The Guardian, 18 December 2024, retrieved 23 December 2024, www.theguardian.com/society/2024/dec/18/hmp-grendon-therapeutic-prison-psychology-offenders-treatment Similarly, Stafford is one of the best performing closed prisons and is a specialist sex offenders’ prison. Prisoners serving time for sexual offences are generally thought to be less violent in prison and more likely to abide by prison rules. ⁵⁹ Institute for Government interviews, 2024. ⁶⁰ BBC, ‘HMP Doncaster sex offender rise ‘to reduce violence’’, 12 October 2017, www.bbc.co.uk/news/uk-england- south-yorkshire-41585901 The same arguments apply to open prisons, which only accept prisoners assessed as a low risk to the public and unlikely to be involved in offending while in prison. Many of these prisoners have already served long periods in prison and are interested in work and training while in custody. And at the other end are reception prisons, which receive a large number of offenders who are regularly in and out of prison and are often only in custody for a short period. These prisoners tend to be younger and are more likely to be violent and less likely to engage in purposeful activity. ⁶¹ Institute for Government interviews, 2024.

…but differences in prisoner characteristics are unlikely to explain all the variation in performance between prisons 

However, these cohort effects are unlikely to explain all the variation in prison performance. There are significant overlaps in prisoner cohorts across most category B, C and open prisons, and these distinctions have only become more blurred as a result of capacity pressure in recent years. Some prisoners have been left struggling to access open or resettlement prisons despite meeting the requirements. ⁶² Independent Monitoring Boards, ‘Written evidence submitted by the National Chair of the Independent Monitoring Boards (FPP0016)’, undated, retrieved 23 December 2024, https://committees.parliament.uk/writtenevidence/125803/pdf Others have been moved more rapidly into open prisons under the temporary presumptive recategorisation scheme, ⁶³ Independent Monitoring Boards, ‘Written evidence submitted by the National Chair of the Independent Monitoring Boards (FPP0016)’, undated, retrieved 23 December 2024, https://committees.parliament.uk/writtenevidence/125803/pdf introduced in March 2023 to move certain category C prisoners automatically to open prisons unless to do so would pose a “wholly unacceptable risk”. ⁶⁴ Inside Time, ‘Prisoners fast-tracked from Seg to open prisons’, 14 August 2023, retrieved 23 December 2024, https://insidetime.org/newsround/prisoners-fast-tracked-from-seg-to-open-prisons/ Despite these changes, violence and protests in open prisons remain much lower than in other prisons. 

Dramatic variation within categories similarly often cannot be entirely explained by differences in the prisoner cohorts. For example, Woodhill is the only category B prison in our list of worst-performing prisons, and one of only two to have received an urgent notification. It is principally a training prison for long-term prisoners, with relatively few new arrivals each week and 95% of prisoners serving a sentence of four years or more, or an indeterminate sentence. This makes the prisoner cohort similar to many other category B prisons such as Dovegate or Gartree, and yet Woodhill has much higher levels of violence, self-harm and drug use, severe staff shortages, low morale and extremely limited purposeful activity. ⁶⁵ HM Chief Inspector of Prisons, Report on an unannounced inspection of HMP Woodhill, 2023, https://cloudplatform- e218f50a4812967ba1215eaecede923f.s3.amazonaws.com/uploads/sites/19/2024/02/Inspection-report-1.pdf These cannot be attributed principally to the prisoner cohort. Warren Hill, by contrast, performs very well and provides a very safe environment, despite holding many high-risk offenders. ⁶⁶ HM Chief Inspector of Prisons, Report on an unannounced inspection of HMP Warren Hill, 2019, https://cloudplatform- e218f50a4812967ba1215eaecede923f.s3.amazonaws.com/uploads/sites/19/2024/05/Warren-Hill-web-2019.pdf

Higher security prisons generally have more officers per prisoner, but there are some extreme outliers

High-security prisons have the highest number of band 3–5 prison officers per 1,000 prisoners, though there is still substantial variation within this group (from 528 to 918).*, ⁶⁸ Ministry of Justice and HM Prison and Probation Service, ‘HM Prison and Probation Service workforce quarterly: March 2024’, 2024, www.gov.uk/government/collections/national-offender-management-service-workforce-statistics  Category B prisons, the next highest security group, also tend to have more officers than other categories, though to a lesser extent. What is particularly striking is how some prisons stand out as major outliers relative to the rest of their category. For category C training prisons, for example, the median number of officers per 1,000 prisoners is 274, while the maximum is 838 – higher than all but one high-security prison. The same is true of the maximum for category B prisons. This shows that a handful of prisons account for a disproportionate share of officers and are therefore much more expensive to run on a per-prisoner basis.

_{* Front-line prison officers are divided into three groups: band 3 and 4 officers, which includes some specialist officers; band 4 supervising officers and band 5 custodial managers. Prison staff numbers are all based on fulltime equivalents, rather than headcount numbers.}