England's rough sleeping snapshot found 4,793 on one night. Its own monthly data found 8,010

On 26 February 2026 the Ministry of Housing, Communities and Local Government published the autumn 2025 rough sleeping snapshot for England. It found 4,793 people sleeping rough on a single night, 3 per cent more than the 4,667 recorded in 2024, 171 per cent above the 1,768 recorded in 2010, and 1 per cent above the previous peak of 4,751 set in 2017. On 20 April 2026, and again on 28 May 2026, the same department published a different set of numbers from the same streets. Its monthly management data recorded 8,010 people seen sleeping rough over the month of December 2025, and 7,541 over March 2026.

Neither figure is wrong. They are answers to different questions, and the gap between them is the most useful thing either one tells you.

The snapshot counts a single night. Each council picks what it judges a typical night between 1 October and 30 November, avoiding weekends, festivals and charity sleep-outs. The official definition covers people bedded down in the open air, or in buildings and places not designed for habitation: doorways, stairwells, tents, sheds. It explicitly excludes hostels, night shelters, campsites, squats and traveller sites. A person who was in a shelter that night was not sleeping rough for the purposes of the statistic, whatever their circumstances the following night.

The method also varies by council. Of the 296 local authorities in the 2025 snapshot, only 51, or 17 per cent, ran a street count. Seventy, or 24 per cent, produced an evidence-based estimate at a multi-agency meeting. The remaining 175, or 59 per cent, held an estimate meeting supplemented by a targeted spotlight count. Homeless Link provides independent quality assurance and graded 93 per cent of councils as assured in 2025, with 17 not fully assured and 3 not assured. The technical report notes that in 2025 the assurance process shifted from extensive in-person verification to remote checks conducted through three phone calls.

The department states the limits plainly. The snapshot is not suitable for estimating the total number of people sleeping rough across a year, nor the total number of homeless people. It records who was seen, or thought to be, out on one night. The technical report concedes that women are systematically undercounted because their rough sleeping is more hidden, transient and intermittent. The 2025 snapshot found 733 women, 15 per cent of the total, with gender unknown for a further 122 people.

The monthly series, the rough sleeping data framework, measures flow instead. Around 98 per cent of English councils report each month, with figures imputed for the rest, and the department calls it more frequent but less robust than the snapshot. In March 2026 the over-the-month figure of 7,541 was 2.2 times the single-night figure of 3,461. Of the March total, 40 per cent had been on the streets long term, 28 per cent were new to rough sleeping, and 13 per cent were returning after a period off the streets.

Interpretation of the snapshot divides. Rick Henderson, chief executive of Homeless Link, called the record high devastating on the day of publication. Writing on 3 March 2026, Emily-Rae Foreman of the Centre for Homelessness Impact argued that the 3 per cent rise was a marked slowdown against annual increases of 20 to 27 per cent in earlier years, that 40 per cent of councils recorded falls, and that London was down 3 per cent to 1,277 people. The Big Issue reported the same day that the single-night method is widely considered an underestimate. The government announced a further 50 million pounds alongside the release.

Australia counts differently again, and has just done so. The Australian Bureau of Statistics derives its homelessness estimate from the Census rather than a street count, and its 2021 estimate of more than 122,000 people, a rate of 48 per 10,000, is dominated by categories England's snapshot excludes by design, including severely crowded dwellings and supported accommodation. Census night fell on 11 August 2026, six days ago, supported by a dedicated homelessness enumeration effort using paper forms and pop-up hubs. Heidi Tucker, chief executive of the Victorian service Anchor, told SBS on 4 August 2026 that not everyone experiencing homelessness is known to a service. Census results are to be released in phases from June 2027.

What is not yet known is whether England's slower increase reflects fewer people on the streets or a changed assurance process, whether the snapshot and the monthly series will ever be reconciled into a single official measure, and how many of the people the Australian Census reached on 11 August will appear in an estimate published almost a year later.