The state counts 69.5
Seventy-five million people are registered with GP practices across the UK, against an official population of 69.5 million. The state's own water planners budget for more people than the census admits, and in 2020 the government quietly stopped counting who had actually left.
Updated June 2026 · UK figures, with England detail
Start with the number that is not in dispute. On the latest NHS figures, 75 million people are registered with a GP in the United Kingdom, five and a half million more than the 69.5 million the Office for National Statistics counts as resident. Both numbers come from the same state, about the same country.
Part of that gap is junk. GP lists carry people who have died or emigrated and were never removed, and the NHS is slowly cleaning them off. So treat 75 million as the number on the books, not a headcount of residents. But "the lists are inflated" stops being reassuring once you notice what sits underneath it: the official count keeps being revised upward, the water industry formally plans for people the census does not see, and the one part of government whose job was to measure who leaves the country gave up in 2020.
In 2019 the ONS and the Home Office jointly decided not to produce an estimate of the irregular population. The Home Office discontinued its exit-check statistics, the figures that tracked whether people on temporary visas actually left, after 2020 and never resumed them. The only number anyone can cite for the UK's undocumented population is a research estimate built on 2017 data: somewhere around 700,000 to 900,000, and now seven years out of date.
ONS already folds small-boat and asylum arrivals into its migration figures, so the Channel crossing is the part the state counts. What escapes the count is quieter: the student-visa holder who stayed, the visitor who never flew home, the worker whose departure was never recorded. The government chose to stop measuring exactly that population, and the gap it leaves surfaces in no headline number.
Water companies have to plan supply for the people who are actually on the ground, so their incentives run the opposite way to a statistics agency's. Under the regulator Ofwat's own reporting rules, every company must add an allowance for "unaccounted for population (clandestine population such as migrant workers, tourists, holiday home owners)" on top of the ONS estimate.
It is a line in their published plans. Thames Water starts from the ONS figure and adds a modelled hidden population of 665,170. United Utilities raised its reported population by several hundred thousand in a single methodology change. Companies sizing the country's water supply plan for more people than the census records, and their plans set down exactly how many more.
The registers, nation by nation, all sit above each nation's official population. England runs furthest ahead; the smaller nations show the same direction at a smaller scale.
Registers are 2026; nation populations are the latest ONS estimates (2024). Against the newer provisional UK figure of 69.5m the gap is about 5.5m; on the 2024 nation breakdown it is nearer 5.9m. Either way it grows, and it is never reconciled.
The honest objection is that this gap is list inflation, and some of it certainly is. But the ONS now builds its own experimental population estimate from this very data, decides the excess is over-coverage, and adjusts it back down to the official figure. Whether that adjustment is right is the whole question, and it rests on a modelling assumption rather than a count. The government has scheduled no census before 2031.
If the official count were simply accurate, its errors would scatter both ways. They do not. The ONS revised net migration for the year to December 2022 up by 139,000, to 745,000. It revised the mid-2023 estimate for England and Wales up by 245,100 when better administrative data arrived. The current 69.5 million headline is itself provisional and, on recent form, more likely to rise than fall when the full figures land in 2026.
A count corrected upward this often has been running behind the population it measures. Other countries have found the same when they finally looked. The United States 2020 census reported a national error of just minus a quarter of a percent, which hid a Hispanic undercount near 5 percent and a Black undercount above 3 percent, both buried under a calm national headline.
Total Managed Expenditure, everything the UK government spends, was about £1,291 billion in 2024-25. The Treasury divides budgets by the official population to plan spending per head. If more people are actually here using services, the same money is spread thinner, and nobody adjusts the budget to match.
Drag the slider between the official count and the number on the books to see how far the money stretches.
At the official 69.5 million, planned spend is about £18,580 per head. The most defensible "people actually here" figure sits between the present-population estimate of roughly 70.3 million and the 75 million on the books. Even the conservative end means services planned for a couple of million fewer people than use them. This measures provision spread thinner, not a clean fiscal loss: uncounted people pay VAT and income tax too, so treat it as dilution rather than a bill.
"The GP lists are just inflated." Partly true, which is why this page does not call 75 million the resident population. It is the number on the books. The sharper fact is that the ONS builds its own admin estimate from these lists, decides the excess is not real, and the government then declines to run a census that would settle it.
"Could it be 80 million?" No. No dataset, physical, administrative or academic, supports ten million people above the official count. The honest ceiling is about 75 million on the books and roughly 70 to 72 million actually present.
"Visitors and tourists aren't residents." Correct, and the headline says "on the books" and "present", not "resident". But the roughly 800,000 visitors here on an average day still use roads, A&E, water and trains. They are a real load on services even when they are someone else's citizens.
"The hidden millions are the small boats." They are not. Channel arrivals almost all claim asylum, and ONS already adds asylum and small-boat flows into its migration figures. The people genuinely missing from the count are visa overstayers and visitors who never left, which is precisely the population the state stopped measuring in 2020.
"The ONS already adjusts for all this." It adjusts, then revises upward year after year, which is the problem rather than the reassurance it is presented as. An estimate corrected up by hundreds of thousands each time better data arrives is not a settled count.
Policy Exchange published The Case for a 2026 Emergency Census in England in December 2025, arguing that migration-driven growth since 2021 makes a mid-cycle count necessary. Centre for Cities called for a "corrections census" after the 2021 base was published. Parliament's own research service has documented the long-standing gap between GP registrations and ONS estimates. The ONS itself is rebuilding population measurement around administrative data because the old method no longer holds. The government's answer: no emergency census, next count in 2031.
GP registers: NHS England Digital, Patients Registered at a GP Practice (England, 63.7m); Public Health Scotland, General Practice Demographics (6.02m); StatsWales / NWSSP, patients registered (~3.31m, Jan 2026); NISRA / BSO, GMS Annual Statistics (2.055m, NI). UK total ≈ 75.0m. Each compared to the latest ONS estimate for that nation.
Official population: ONS, Provisional UK population estimate, mid-2025, 69,487,000, published 27 November 2025. Usual residents, 12-month definition.
Present population: ONS usual residents plus an average daily visitor figure derived from VisitBritain inbound nights (293 million in 2024, ÷ 365 ≈ 800,000). A derivation, not an official total.
Water "clandestine population": Ofwat RAG 4.11 reporting guidance; Thames Water and United Utilities WRMP24 demand methodologies. Company "population served" starts from the ONS estimate and adds a modelled uplift, so it is not an independent count; it is presented here as the industry's own stated belief that the census undercounts.
Irregular population: Pew Research Center, unauthorised UK population, 700,000–900,000 (2017 basis; the earlier 1.2m figure was retracted in March 2025). Exit-check discontinuation per Home Office immigration statistics.
Caveats: GP registers include genuine list inflation, duplicates and delayed de-registration; the 75m is "on the books", not a resident headcount. Visitor and short-term figures are presence, not residence. The defensible claim is not an exact hidden total but that several independent systems sit above the official count, the state's own planners assume an undercount, and the government has stopped measuring the population that would close the gap.