Mistaken Releases from Prison

In recent months, the mistaken release of prisoners has been headline news in the UK. But how often does this happen, and why? This article will discuss the data behind the headlines and the underlying issues in the criminal justice system leading to these errors. 

Ministry of Justice Data on Prisoners Released in Error

The Ministry of Justice gathers and publishes data across four categories of prisoners being ‘unlawfully at large’.  Touching briefly on each of these categories:

• 1. ‘Escape’ is defined as overcoming a physical security measure to leave lawful custody, either prison, Court or transportation.

• 2. ‘Absconding’, meanwhile, is defined as an escape from open prison conditions, meaning that no physical security measures are breached.

In the year to March 2025, there were 12 escapes, all of which occurred during transportation, and 57 absconds. 2 escapees and 15 absconders remained at large after 30 days. 

By far the largest data category of being unlawfully at large is:

• 3. ‘Temporary Licence Failures’, with 863 failures in the year to March 2025.

A Release on Temporary Licence (often referred to as 'ROTL' - pronounced 'rottle') may be granted to a prisoner to enable them to leave custody for a short and specified period of time and for an approved purpose. A ‘Temporary Licence Failure’, therefore, includes a breach of any condition attached to a temporary license such as where the person may go, who they may meet, and the time at which they may return. 44 of this category were failures to return to prison at all; 12 of which were still at large after 30 days. 

• 4. ‘Releases in error’, however, is the category which has gained most media attention in recent months and also represents the most meaningful increase from the previous year. A release in error refers to a prisoner being released from a Court or prison when they are not supposed to be. The figure in the year to March 2025 was 262, an increase of 128% from the previous year.

Interestingly, the House of Commons Library published an article on 12 November 2025 stating that this figure included instances of prisoners being released too late, as well as too early, but this claim is not repeated or reflected in the Ministry of Justice data.

Why are Prisoners Released in Error?

Recent high-profile instances of early release in error did little to boost confidence in the criminal justice system, not least because of some of the circumstances reported upon. In one case, the cause of the error was given as the court incorrectly marking a 45-month sentence as being a suspended sentence, which is not a sentence presently available in law. It was reported that the error was spotted and amended by the court, but the correction was then sent to the wrong person, and the prisoner was released from the prison (the article can be found here). 88.9% of prisoners released in error were released from prisons rather than courts, but the underlying error causing release may originate from the court or the prison.

There have been a variety of reasons suggested for the increase in release in errors, in particular the fact that the system is stretched and underfunded, and the implementation of new early release schemes.

In his statement in Parliament on 27 October 2025, the Justice Secretary drew particular attention to reductions of 24% in funding and 30% in staffing for prisons between 2010 and 2015 and commented that 50% of frontline prison officers have less than 5 years’ experience. However, HM Chief Inspector of Prisons noted that this was more than a problem of inexperienced individuals, describing the issue as ‘endemic’ and ‘structural’.

Indeed, a report by HM Prison Inspectorate on HMP Pentonville published in July 2025 referred to the ‘backlogs in sentence calculations’ leading to releases in error as well as unlawful detention. This means that even at the point someone is in prison, serving their sentence, the prison itself may not know the exact length of that sentence. It is easy to see how this may lead to unlawful detention, particularly for short sentences or where significant time has already been served. It is perhaps more difficult to understand how this leads to early releases but is certainly a sign of a system under pressure.

Meanwhile, one example of a new early release scheme caused a demonstrable rise in early releases in error. The Criminal Justice Act 2003 (Requisite and Minimum Custodial Periods) Order 2024 changed the release point for standard determinate custodial sentences from 50% to 40% beginning in September 2024.  Specified offences were excluded from the scheme including offences of domestic violence, which in turn included the offence of breaching a restraining order under the Sentencing Act 2020.  However, it did not include the offence of breaching a restraining order under the Protection from Harassment Act 1997.  This omission was an error and led to the release of 37 prisoners who had been convicted of an offence under the 1997 Act. This single issue accounts for 14% of releases which were then classified as being “in error” in the year to March 2025.  Even if these are removed from the data, the increase in releases in error would still be 95.6% from the year to March 2024, almost doubling.

The data for the year to March 2025 also notes that, for the previous year, only 75.6% of the instances of release in error were identified within the reporting period, with the data being updated later. The data from the year to March 2024 does not show or refer to the updated figure. This also leaves open the possibility that there are further releases in error for the year to March 2025 not included in the figure currently published, albeit that updates may have already been made.

An independent review is due to report in February 2026 on the causes of release in error as well as the accuracy and transparency of data on releases in error. This may shed further light on the issues discussed in this article, and more detailed data could lead to clearer conclusions being draw. We will provide a further update upon the publication of this report.

The information in this article is accurate as at the date of publication. For further information, please contact our team.

Read more about the author, Connor Byrne here.