England and Wales release the first 700 under the one-third rule

About 700 prisoners walked out of jails in England and Wales on the morning of 1 October, the first wave of a permanent early-release scheme written into the Sentencing Act. The Ministry of Justice expects roughly 4,500 people to leave before their previous dates, in ten waves running to June 2027. About 2,550 of them are due out by the end of this year. Spread evenly, Thursday's 700 would be about six people per prison. They will not be spread evenly. The releases track sentence length, not geography.

The rule change is specific. Prisoners who had expected to serve half their term in custody and half on licence will now leave at the one-third point, if they qualify. The first eligible group are those jailed for up to 19 months. From November the net widens to sentences of up to three years and two months. By next June, people on fixed terms of more than 12 years, who had expected to serve two-thirds inside, become eligible to move to a community licence at the halfway point. The total sentence does not shrink. The custody portion does. The scheme applies, in principle, to about 70,000 offenders. About 18,000 are excluded outright.

The exclusions are the part Prime Minister Andy Burnham tightened before the Act took effect. Murder stays out. So does manslaughter, in all its forms. Sexual offences and grooming stay out. So does causing or allowing the death of a child, and a driving offence that led to a death. Burnham's review cut the expected release pool from about 6,000 to about 4,500. Flights and paperwork already booked under the old Channel returns deal are a separate matter. This scheme is domestic, and new cases under the excluded headings do not enter it.

The capacity number that forced the timetable is public. Ministry of Justice data showed 87,499 people incarcerated as of Monday, 98.2 percent of available space. Officials have said the estate is months from being full. That is why a permanent change to release points replaced the emergency schemes of the last two years. Between October 2023 and September 2024, a Conservative early-release programme freed more than 10,000 people. Labour's version, now Burnham's, is smaller per wave and written into statute rather than activated by a crisis order. It is also slower: ten waves over nine months, instead of a single dump.

Who left on Thursday is narrower than the headline. The first wave is the short-sentence group, up to 19 months, leaving at one-third rather than one-half. A person sentenced to 18 months who expected nine months inside is now eligible at six months, if not excluded. The BBC's account of the scheme says the first 700 were due out on 1 October if the eligibility checks cleared. Probation has to absorb them on licence. The Act does not add a published figure for extra probation staff tied to Thursday's wave. That is the operational gap. Custody pressure is being moved onto a supervision system whose caseload was already the reason some of these sentences were imposed in the first place.

The politics sit in the exclusions. Burnham's intervention, taking manslaughter, grooming and child-death offences out, was a response to the charge that a capacity measure would cover serious violence. The remaining pool is still people convicted of imprisonable offences. Victims' groups will look at the one-third date, not at the list of offences that were carved out. The Conservative scheme's 10,000 releases are the comparison opponents will use. Labour's answer is the smaller number, the statutory exclusions, and the claim that 98.2 percent occupancy leaves no other lawful option that keeps the estate inside its certified capacity.

The next dates are already set. A second wave follows in November, when the sentence ceiling rises to three years and two months. The last of the ten waves is slated for June 2027, when the long fixed-term group moves from two-thirds to one-half. Between those points the estate has to show that 4,500 early releases actually hold the population below the ceiling, or the same argument will return with a larger number attached. The open question is the recall rate. If a material share of Thursday's 700 come back on breach within months, the capacity gain is borrowed, not banked.