Prison Officer Assault Statistics 2026: Official Figures

In shortThere were 9,833 recorded assaults on staff in the 12 months to March 2026, 7% lower than the year before but still well above pre-2023 levels.

• Assaults on staff9,83312 months to March 2026; provisional
• Rate113 per 1,000Prisoners
• Serious assaults871A rate of 10 per 1,000
• Per day (our arithmetic)About 27Of which about 2.4 serious

The figures come from the Ministry of Justice's Safety in Custody quarterly update, published on 30 July 2026, summary Tables 4 and 7, and the March 2026 figures are provisional.

The direction is more nuanced than either 'rising' or 'falling'. So the latest year is better than the one before it, but it remains well above the levels recorded before 2023 and above the 2016 baseline.

• The number fell by 7% in the latest year.
• Before that it rose from 7,461 in the year to March 2023 to 9,847 in the year to March 2024 and 10,568 in the year to March 2025 (revised).
• The latest quarter, January to March 2026, was 8% lower than the previous quarter at 2,242.

In shortAssaults on staff are 82% above 2016 and 40% above the pandemic low of 7,011, but serious assaults fell as a share of the total.

• Year to March 20165,409
• Pandemic low, year to March 20217,011
• Peak, year to March 2025 (revised)10,568
• Year to March 20269,83340% above 2021, 82% above 2016 (our arithmetic)

The table gives the annual series for every 12 months to March from 2016 to 2026.

Assaults on staff rose from 5,409 in the year to March 2016 to 9,630 in the year to March 2019, fell to 7,011 in the year to March 2021 when prisons were operating under pandemic restrictions, and then rose again to 10,568 in the year to March 2025 before falling to 9,833.

Serious assaults on staff peaked at 947 in the year to March 2024 and were 871 in the year to March 2026.

They fell as a share of all assaults on staff from 11.9% in the year to March 2016 (644 of 5,409) to 8.9% in the year to March 2026 (871 of 9,833), by our arithmetic. That means the growth in the headline number came mainly from assaults that were not classed as serious.

In shortAssaults on staff have fallen for several quarters to 2,242 in January to March 2026, which is encouraging but not yet a settled trend.

• Latest quarter, Jan to Mar 20262,24212.6% lower than a year earlier (2,565)
• Highest quarter in the table (from 2003)2,715July to September 2024; latest is 17.4% lower
• Serious assaults, Jan to Mar 2026188267 in January to March 2024

Quarterly figures show the recent turn more clearly than the annual table.

Assaults on staff were 2,602 in January to March 2024, rose to 2,715 in July to September 2024, stayed between 2,565 and 2,572 through the first three quarters of 2025, and then fell to 2,447 in October to December 2025 and 2,242 in January to March 2026.

Serious assaults on staff follow a broadly similar path: between 216 and 248 in the following quarters of 2024 and between 224 and 232 in 2025. The four most recent quarters sum to 871, which matches the annual figure.

In shortThe female estate recorded a new peak of 311 assaults on staff per 1,000 prisoners, against 105 in male establishments.

The women's estate is small, so a modest number of incidents moves the rate a lot. Female establishments held an average of about 3,490 prisoners in the year, which is 4.0% of the 87,283 total, yet they recorded 11.0% of assaults on staff (1,085 of 9,833), by our arithmetic.

Serious assaults on staff in female establishments were 53 (15 per 1,000), 15% lower in rate than the year before. The publication also records that the overall assault rate in female establishments was 660 per 1,000, a peak, compared with 342 in male establishments.

The data do not explain why. We would not infer causes from the numbers, and the MoJ does not do so in the tables.

In shortAssaults on staff, at 9,833, are about 32% of the 30,820 assault incidents recorded, and other data and union surveys describe further pressures.

• All assault incidents30,820354 per 1,000 prisoners
• Prisoner-on-prisoner assaults21,245244 per 1,000, up from 237
• Serious assaults of all kinds3,41939 per 1,000
• Assaults on staff shareAbout 32%Our arithmetic; a subset, and some incidents are recorded in both categories

Assaults on staff are one part of a larger picture. In the same year the MoJ recorded the incident totals shown above.

HMPPS Annual Digest 2025-26

Reports 26,630 drug finds, 13,284 weapon finds and 11,916 mobile phone finds in prisons, along with 14 escapes, and says too few prisons met random drug testing levels to produce a reliable prevalence estimate. These figures are context rather than evidence of cause. We do not claim that any of them explains the assault trend.

POA 2026 member survey (reported by the PSPRB)

61% of its 1,600 respondents said they had been assaulted while at work and around 40% did not feel safe in their workplaces. The PGA's evidence also described assaults and self-harm as at record highs. These are union survey results.

PSPRB on sickness absence

Sickness absence among band 3 to 5 officers is high, with mental ill health the largest recorded reason (see the staffing and attrition guide). We read these sources side by side and do not weigh them against each other.

In shortWe offer no explanation for the rise to 2025 or the fall since, because we could not find an official analysis that isolates a cause.

The MoJ's notes are explicit that year-to-year rises and falls may not reflect underlying trends because prisons change: a new wing opens, a block closes for refurbishment, a prison changes role, or the operating model changes.

Table 10 lists major openings, closures and re-roles since 1997, including Dartmoor's temporary closure in September 2024, Cookham Wood's change from a youth to an adult male prison in July 2024 and Lowdham Grange's return to public management in August 2024. Any of these can shift counts at establishment and national level.

• Because rates are per 1,000 prisoners, a falling count with a stable population gives a lower rate, and a rising population with a stable count gives a lower rate too.
• The national population average was about 87,000 in both of the last two years, so the 7% fall in the number and rate in the latest year reflects a real fall in recorded incidents rather than a denominator effect.
• By establishment, many small cells in Table 8e are suppressed, which makes per-prison annual sums unreliable, and we do not use them.

In shortGet medical attention, report it internally and to the police, and note the union's time limits: two years for CICA and a three-year personal injury limit.

This section is general information, not legal advice. If you are assaulted at work, get medical attention first. Report the incident through your establishment's procedures and tell your line manager and, if you are a union member, your branch official.

Report it to the police: the POA's legal handbook says all members have a right to refer assaults to the police and that failure to report an incident to the police as soon as is reasonably practicable can disqualify a Criminal Injuries Compensation Authority claim.

1. Get medical attention and make sure you are safe.
2. Report the incident through HMPPS procedures and to your line manager; make sure it is entered in the accident book.
3. Report it to the police and keep the crime reference number.
4. If you are a union member, tell your branch official and check the scheme time limits (CICA two years; personal injury report within two years six months, per the POA handbook).
5. Keep records: incident report, photographs, medical notes, witness names, time off and expenses.
6. Ask about civil service injury benefit and sick pay before your sick pay runs out.

In shortUse the national figures for national statements, not to judge how safe one prison is, and treat the items below as unverified.

HM Inspectorate of Prisons reports are the better source for a single establishment, and the prison finder links to the latest report where we have one.

We could not verify several things:

• The figures for the year to March 2026 are provisional and may be revised.
• We did not analyse per-prison counts.
• We did not find an official breakdown of assaults on staff by the type of staff member, by grade or by whether an officer was injured, in the summary tables we read.
• We did not verify the POA and PGA survey statistics beyond how the PSPRB reports them.
• We do not state causes for the trend.