Premier League Viewership Statistics: the 2025/26 Season in Numbers

Nobody publishes a verified count of how many individual people watched a live Premier League match somewhere in the world during 2025/26. The league does not produce one, and no measurement body compiles a single global viewing figure across nearly two hundred separate television markets.

What the Premier League does publish are three reach measures, and they are frequently quoted as though they were viewing figures. In its own end-of-season summary the league said coverage was distributed to 189 countries out of 193 United Nations member states, that 900 million homes were able to watch, and that 1.87 billion people engage with Premier League media at least weekly (Premier League season review). The league has also said that 34% of those followers became interested within the previous four years, which describes the composition of the following rather than growth in match viewing.

None of those three numbers counts people watching football. A home able to receive coverage may never switch it on. Weekly media engagement includes reading a score, watching a clip or following a club account. They are useful measures of the league’s commercial footprint, and they are the right figures to cite for that purpose — but a headline claiming that “1.87 billion people watch the Premier League” misstates what the league actually said.

Older reach claims still circulate widely. Figures such as a 4.7 billion potential worldwide audience, or 800 million homes across 188 countries, come from research more than a decade old and are repeated without their original methodology. They are recorded in the workbook’s Excluded_or_Unverified sheet with the reason for rejection, rather than quietly dropped.

The 2025/26 season was the first of a four-year domestic rights cycle that changed how much Premier League football a UK viewer could watch. Sky Sports moved from 128 exclusively live matches a season to a contracted minimum of 215, taking on every displaced Sunday 2pm kick-off, the returning Sunday 4.30pm slot, Friday and Monday night matches, three midweek rounds and all ten matches on the final day. TNT Sports kept 52 matches. Amazon Prime Video, which had shown 20, did not bid for live rights in the new cycle.

That matters for reading any UK figure from this season. Adding roughly a hundred matches to a broadcaster’s schedule raises its total viewing almost automatically, whatever happens to the size of individual audiences. The two measures have to be kept apart.

The final day gave the clearest single reading. On 24 May 2026 Sky broadcast all ten matches live at once for the first time in the UK, and the coverage averaged more than three million in-home viewers — a figure that excludes pubs, clubs and every other out-of-home setting. Arsenal had secured the title a week earlier, with Tottenham Hotspur’s survival and the last European places still unresolved.

Sky was also the exclusive home of the run-in: every fixture from 5.30pm on 9 May was shown only on Sky. Competitive tension appears to have helped. The 2024/25 title race was effectively over by April, and UK viewing fell that season; 2025/26 went to the penultimate week and produced a third different champion in three years.

One caution about causation: nobody has isolated how much of the 2025/26 increase came from a closer title race, how much from the larger schedule, and how much from the reshaped slot structure. Those explanations are plausible, and they are what the broadcasters themselves point to, but they have not been tested.

Percentage changes are only half a picture. The obvious follow-up question — how many people watched an average Premier League match on Sky in 2025/26 — does not have a published answer. Sky released the direction of travel but not the levels, and TNT Sports published nothing comparable for the season at all.

The most recent itemised figures by broadcaster date from 2024/25, and they were reported from data circulated to Premier League clubs rather than from a published dataset. On that basis Sky’s average match audience fell about 10% to 1.57 million, from 1.78 million in 2023/24, while TNT Sports averaged 1.15 million, down 5% for its Saturday lunchtime matches and 17% once midweek fixtures are included. Those numbers are shown here because they are the only itemised broadcaster levels available, and flagged throughout as unverified against a primary source.

The BBC’s position is different again, because it holds highlights rather than live rights. From the start of 2025/26 its contract allowed it to publish clips of every match from 8pm, rather than waiting to put the whole Match of the Day programme on iPlayer at 10.30pm. By January 2026 the BBC reported that Match of the Day content was averaging 1.52 million weekly views across iPlayer, the BBC Sport app and the website, against a loss of 770,000 television viewers over the previous twelve months.

Those two figures are widely quoted as though the digital gain covers the television loss twice over. It does not follow. A view is a play of a piece of content; a television viewer is a person. One person watching four clips generates four views. The two measures are not subtractable, which is why they appear in separate panels below.

The United States is the market with the fullest published record, because NBCUniversal releases a detailed season summary every May. For 2025/26 it reported that matches across NBC and Versant’s USA Network averaged 535,000 viewers per Premier League television match window, up 5% on 2024/25 and the second-highest US season average recorded, behind only the 546,000 of 2023/24. It was also the fifth consecutive season in which the English-language average topped 500,000.

Beneath the headline, the shift towards the main broadcast network was sharper. NBC itself averaged a Total Audience Delivery of 1.2 million viewers, its most-watched Premier League season, and 18 NBC matches averaged at least a million — half as many again as the 12 recorded a year earlier. Total consumption across NBCUniversal platforms in English and Spanish reached 18.33 billion minutes, up 7%.

The season opened and closed on records. The opening weekend of 15 to 17 August 2025 averaged a TAD of 850,000 across six matches, up 4% on the previous record, and closed with Championship Sunday on 24 May 2026: all ten matches shown simultaneously across NBC, Peacock, NBCSN, USA Network, CNBC, SyFy, Telemundo, Universo and NBC Sports digital platforms, averaging a combined 2.4 million viewers in English and Spanish. The Tottenham Hotspur against Everton finale alone averaged 1.1 million on NBC and Peacock, the most-watched English-language final-day match on record.

Over the longer run, NBCUniversal says that since taking over US coverage in August 2013 it has presented 48 of the 50 — and 98 of the 100 — most-watched live Premier League matches in the country. Four of the top 20 came in 2025/26 alone.

Individual-match records are kept market by market, and this report keeps them that way. A UK audience measured by BARB and a US audience measured by Nielsen across broadcast, cable and streaming are not the same quantity, and merging them into a single worldwide ranking would produce a list that means nothing.

The United States has a published, ranked list from the rights holder. Manchester City’s 2-1 win over Arsenal on 19 April 2026 — second against first, with six points between them and eight matches to play — averaged 2.6 million viewers across NBC, Peacock, Telemundo and NBC Sports digital platforms, and stands top of it.

For the UK, no equivalent ranked list of individual-match audiences was published for 2025/26. The closest verified figures are the more than three million in-home viewers for Sky’s simultaneous final-day coverage, and the Premier League’s own count from 2023/24, when ten live matches drew UK audiences above three million — the highest number on record at the time. That count has not been updated since.

The clearest platform shift in 2025/26 was structural rather than behavioural. With up to four Premier League matches kicking off at 2pm on the busiest Sundays, Sky launched Multiview, a single stream that cuts between every concurrent match with its own commentary and half-time analysis, available through Sky TV platforms, NOW and the Sky Sports app. Individual matches remained available on separate channels alongside it. It is a direct product response to a schedule that now regularly splits the audience.

In the United States, Peacock accounted for two of the seven most-streamed Premier League matches on record during the season. In 2024/25, the last season for which a figure was published, more than four billion minutes of Premier League matches were streamed on NBCUniversal platforms — the second consecutive season above that mark, and a subset of the 17.14 billion minutes consumed in total.

Clips are not matches

Social and short-form numbers are the easiest to misread. NBC Sports reported a four-year high in Premier League social impressions and video views in 2025/26, with impressions up 17%, video views up 24% and sharing up 42%, while its YouTube channel passed 180 million views for a fourth consecutive season and accumulated 1.3 billion minutes watched.

Those are highlight and clip figures, counted on entirely different rules from television audiences: a play of a 40-second goal clip counts once, as does 90 minutes of live football. They belong in a separate column from match audiences, and they are kept in one throughout this report and the accompanying dataset.

Out-of-home viewing

Out-of-home viewing — in pubs, clubs and hotels — is systematically absent from the headline UK figures. Sky’s final-day average of more than three million is explicitly in-home, and the true total for that afternoon was higher by an amount nobody has published.

What exists instead is survey evidence. Ipsos research for Sky Business found that 70% of pub-going sports viewers agreed the increased number of Premier League matches gave them more reasons to visit hospitality venues, and 63% said they had watched more Premier League matches in pubs and bars than the previous season. That is stated behaviour rather than measured viewing, and it should be read as such: no verified out-of-home audience figure for the 2025/26 Premier League season was published by the research cut-off.

Piracy

Illegal streaming is regularly cited as a driver of falling audiences. This report does not quantify it. The figures still circulating date from the 2019/20 season and rest on self-reported survey answers with no published methodology; nothing recent and methodologically credible was found for 2025/26. Rather than repeat an old number, the gap is recorded in the workbook’s Excluded_or_Unverified sheet.

Outside the UK and the US, the Premier League’s footprint is well documented and its audiences are not. The league publishes which broadcaster holds rights in each territory, but no comparable audience figures for 2025/26 were published for India, China, sub-Saharan Africa, the Middle East, Australia, Japan, South Korea or south-east Asia by the research cut-off, and most rights holders in those markets do not release season-level viewing data at all.

Rather than build a comparison from figures that would not be comparable, this section records what can be verified: who carries the coverage, and on what kind of platform. The pattern worth noting is the shift towards streaming-first distribution — Migu in China, Coupang Play in South Korea, U-Next in Japan, Stan Sport in Australia and JioStar in South Asia are all digital-led, which changes both how matches are watched and how audiences are measured, or whether they are measured publicly at all.

Rights deals are not audience figures, but in the UK they set the ceiling on what any audience figure can be. The number of matches a viewer is allowed to watch changed sharply in August 2025, and every UK comparison across that boundary carries the change with it.

One question the new cycle was expected to answer — whether flooding the schedule would dilute audiences — got a provisional answer of no. Sky added roughly a hundred matches and its average match-slot audience still rose 9%. That is one season, in a year with an unusually close title race, and Sky did not publish the absolute levels that would let anyone check the size of the effect. It is a data point, not a settled finding.

Five gaps are worth stating plainly, because they shape what can honestly be claimed about Premier League viewership in 2026.

There is no verified global viewing figure. The reach measures the league publishes describe availability and following. Any single worldwide viewing number in circulation is either a cumulative total that counts people repeatedly, or an estimate whose methodology cannot be inspected.

UK absolute levels for 2025/26 are missing. Sky published percentage changes; it did not publish the average audiences behind them. TNT Sports published nothing comparable. The most recent itemised broadcaster levels are from 2024/25 and come from figures circulated to clubs rather than a released dataset.

Out-of-home viewing is largely uncounted. Headline UK figures are in-home. Pub and hospitality viewing is real and, on the evidence of Sky’s own survey work, growing — but it is not in the numbers, and no measured out-of-home audience was published for the season.

US comparisons cross a measurement change. Nielsen Big Data + Panel became the currency in autumn 2025. Every US year-on-year comparison involving 2025/26 spans that change, and the direction of the bias is known: the new system generally measures more people.

Most international markets publish nothing. For India, China, Africa, the Middle East and most of Asia and Oceania, distribution is documented and audiences are not. Comparative international tables built from such markets are almost always assembled from incompatible or undated sources.

Research cut-off: 4 August 2026. Every source was accessed and checked on that date. The 2025/26 season — 15 August 2025 to 24 May 2026 — is the main reference period.

Source hierarchy. Rights holders, the Premier League and audience-measurement bodies came first; specialist broadcasting and media-rights publications second; general news organisations only where they identified their underlying data source. Where several outlets repeated one statistic, the original issuer was cited rather than the repetitions. Search-result snippets, AI-generated summaries, content aggregators, encyclopaedias and figures copied between statistics articles without checking the original were not used as evidence.

Verification. Every numerical claim carries a source name, a direct URL, a publication date, an access date, the relevant season, the market, the platform or broadcaster, the metric definition and its unit. Rounded source values were kept as published rather than converted into false precision, and the original wording of each figure is preserved in the workbook.

Calculations. Percentage changes use ((new value − previous value) ÷ previous value) × 100 and were recalculated independently from the underlying values; where a rights holder had published its own rounded percentage, the two agreed. Calculated results are labelled “Calculated by SportingPedia” and are never presented as official broadcaster or league figures. No percentage change was calculated where the previous value was missing, where the two values used different definitions, or where a methodology change made the comparison meaningless without qualification.

What was rejected. Figures that could not be traced to an original issuer, that were superseded, that were contradicted by a primary source, or whose methodology could not be inspected were excluded and logged in the workbook’s Excluded_or_Unverified sheet, with the reason for each. Nothing was dropped silently.

Note on illustrations

The previous version of this page carried four images. Three were illustrations recycled from unrelated articles — an NFL betting explainer from February 2021, a Ligue 1 betting overview from August 2022 and a gambling illustration from January 2021 — each given alt text copied from the section heading rather than describing the image. None depicts Premier League viewership, and reusing them here would have meant labelling artwork as something it is not. The fourth was a hero graphic dated 2023, which cannot be presented as current.

None of the four has been reused. Every visual on this page is a newly generated inline SVG built directly from the verified data in the accompanying workbook, so the page remains a single self-contained file with no external image dependencies. The original asset URLs are preserved in the workbook’s Asset_Audit sheet, together with what each one actually shows and why it was not used.