How Many Americans Will Bet on the NFL in 2026? The Honest Answer

The number that got retired

For a while there was an answer. Every summer the American Gaming Association, the casino industry's trade group, commissioned a survey and announced how many Americans planned to bet on the coming NFL season. In 2022 it was 46.6 million adults, 18% of the country. In 2023 it was roughly 73.5 million, 28%. Then it stopped.

Look at those two numbers again. Twelve months apart, the second is 57.7% higher than the first. A population does not move like that in a year, and no published source explains why this one appears to. What is wobbling is more likely the instrument than the country.

The instrument was also measuring something looser than most readers assumed. As ESPN reported at the time, the survey's definition of a bet included pick'em and squares pools, paid fantasy contests, and casual wagers with family or friends. Even the 2022 release, which led with 46.6 million, noted inside the same announcement that 23 million people planned to bet online and 6 million in person at a sportsbook. In that release the headline was far larger than either sportsbook channel. For the 2025 season the AGA published a dollar figure instead—an estimated $30 billion wagered legally, and stopped counting people.

One survey, four answers

The clearest illustration of the problem is a single AGA release from February 2024, about a single game. Before Super Bowl LVIII, the association reported that 67.8 million American adults—26%—were expected to bet on it. The same release then broke that down. 42.7 million planned a "traditional" wager online, at a retail sportsbook, or with a bookie. 36.5 million planned to bet casually with friends or in a pool or squares contest. And 28.7 million—11%—intended to place an online wager with a legal U.S. sportsbook.

The answer moves by 39 million people depending only on where the line is drawn. Nor can the sub-categories be added up: the middle two alone come to 79.2 million, well above the 67.8 million headline, because plenty of people intended to do both.

A year later the AGA changed methods, and to its credit said so plainly: "Previously, the AGA, using a public survey instrument, arrived at estimates that incorporated all forms of betting on the Super Bowl, including bets made on unregulated or illegal platforms, as well as casual bets (squares, etc.)." Since then it has estimated legal wagers only. Before Super Bowl LX it projected $1.76 billion. That release contains no count of people anywhere.

Two surveys, the same number, different people

For sports betting generally, the most defensible figure available comes from Pew Research Center: 22% of U.S. adults said they had bet money on sports in the past 12 months. That rests on a probability panel of 9,916 adults recruited by home address, fielded in July and August 2025, with a margin of error of 1.3 points.

But the 22% is a net across three separately asked items. Fifteen percent said they had bet with friends or family—a private pool, a fantasy league, a casual bet. Ten percent said they had bet online with an app, a sportsbook, or a casino. Eight percent said in person. Those groups overlap, so they do not add to 22.

So one survey supports two very different sentences: twenty-two percent of American adults bet on sports last year, and ten percent used an app or a sportsbook. Both are accurate. Only one describes the thing most people picture.

Now the coincidence. The Siena College Research Institute and St. Bonaventure University also report that 22% of Americans are "bettors"—from an entirely different instrument. Theirs is a weighted opt-in online panel of 3,084 U.S. residents fielded in February 2026, and the release is candid about what that means, noting that "the probability of being included in any given online survey sample is unknown, very difficult to ascertain, or simply zero." Its 22% is built from two questions: 27% say they currently hold an online sportsbook account, and 83% of those say they have ever placed a bet through it.

“

They share a number and almost nothing else.

So one 22% is past-year betting including the office pool. The other is people who hold an account today and have bet through it at some point in their lives. Neither is wrong. They are simply not the same measurement, and they should never be cited as corroborating each other.

Then there is how the question gets asked. In 2026 Gallup ran its gambling questions two ways in overlapping weeks—by telephone and through its probability-based web panel. Across its six sports-betting items the phone survey came out at 15% and the web panel at 21%. Gallup's own explanation is that phone respondents may be reluctant to tell a live interviewer they gamble. Same organisation, same questions, same summer, six points apart.

Dollars are not people

Americans legally wagered $166.94 billion on sports in 2025, by the AGA's compilation of state regulator reports for commercial operators—tribal sportsbooks are not in it. That is a real number, and it is not a headcount. Handle counts the same money every time it goes back into play: a single $100 bankroll recycled ten times is $1,000 of handle.

That NFL-specific $30 billion is an estimate resting on an estimate, arrived at—in the AGA's own description—by applying national handle growth "to last year's estimated NFL season total, along with football-specific reporting from select states." It does not name those states. Its window runs from preseason through futures booked as early as March, the playoffs, and the Super Bowl.

Even where regulators publish sport-level data, NFL exposure cannot be isolated. Colorado is one of the few states that breaks out wagers by sport: in September 2025 pro football was about 32% of the state's handle, while parlays sat in a separate bucket of about 27%—and a parlay can contain football legs. The football line is a floor, not a total.

Dollar figures also disagree with each other by scope. The AGA's Super Bowl LX projection was a national record. When the game was over, Nevada's regulator reported $133.8 million wagered across the state's 186 sportsbooks—the lowest of the ten Super Bowls in its own table. One is a national projection of the legal market; the other is what one state's books actually took. Both can be true because they are counting different things.

Accounts are not people

The operators publish customer numbers, and the two largest both tell you, in their filings, not to read them as people. DraftKings reported an average of 4.0 million "monthly unique payers" for 2025. Its own definition is unique paid users per month "across one or more of our Sportsbook, iGaming, DFS, digital lottery courier, prediction markets or other product offerings via our technology." So it folds in online casino, daily fantasy, a lottery-courier app and prediction markets; it is an average of twelve monthly snapshots rather than a count of people across a year; and it spans 27 U.S. jurisdictions plus Ontario, Canada.

Flutter, FanDuel's parent, is blunter still. Its 2025 annual report says the player metric "reflects a level of duplication that arises from individuals who use multiple brands," and then, in the next sentence: "we are generally unable to identify when the same individual player is using multiple brands, and we therefore count this player multiple times."

Flutter also says it cannot quantify that duplication, does not believe it material, and locates it mainly in its UK and Ireland brands. Even with the qualifier, a filer has told its regulator that a published player count is a count of accounts. Adding it to a competitor's figure would compound the problem rather than solve it—neither company can see the other's customers, neither number is confined to sportsbook betting, and neither is confined to the United States.

A wager is not a bettor

One widely circulated NFL-specific dataset comes from Optimove, a marketing-technology vendor whose customers are betting operators. Across the 2025/26 season, from preseason through the Super Bowl, it observed 136,211,778 wagers from 5,047,205 bettor records.

It does not say how many operators supplied that data. Or which ones. Or what countries they operate in. Or whether someone holding accounts at two of them is counted once or twice.

Without those facts the figure has no denominator. It is a large window onto people who already bet—not a sample of Americans, and not something that can be scaled up into one. It also works out to roughly 27 wagers per record, which is its own reminder that counting wagers and counting bettors are different jobs.

The same report carries a separate survey of 425 NFL bettors aged 21 and over in households earning at least $75,000. Optimove is explicit that "the two datasets do not represent the same individuals"—worth holding onto, because the two disagree. In the survey, 37% said they would bet the Super Bowl. Among the observed bettor records, 21% did.

Four questions to ask any NFL betting number

We went looking for a public dataset that counts NFL bettors as people. We did not find one. Regulators publish dollars. Operators publish their own payers and players. Vendors publish wagers. Nobody publishes people.

So when a number reaches you this season—and one will—put four questions to it before repeating it.

• Is it people or dollars? Handle, revenue and wager counts are money and events, not persons.
• Is it measured or projected? A figure published before the game is a forecast, not a result.
• Does it cover only legal, regulated wagering? Prediction-market contracts on NFL games sit outside every state's handle report—a point the AGA itself makes, noting that these markets operate entirely outside state sports betting regulatory frameworks.
• What counted as a bet? A sportsbook account, or the squares grid in the office kitchen?

Keep one more thing in view, because the biggest numbers travel furthest. The National Council on Problem Gambling's own survey found that most Americans did not place a sports bet in the twelve months before it asked in early 2024. Gallup describes betting on college and professional sports as relatively flat since 2003 in its telephone series. And before Super Bowl LX, YouGov asked 3,023 adults whether they planned to make any bets related to the game; 79% said they did not.

How many people bet is also a separate question from how many people are harmed, and the second does not follow from the first. The NCPG says of its own risk findings that it "cannot determine the degree to which these factors cause problem gambling"—sports betting participation among them. Treat any number that slides from one to the other with suspicion.

• If gambling is becoming hard to control in the United States, the National Council on Problem Gambling lists confidential help and state resources. If you may hurt yourself or are in immediate emotional crisis, call or text 988 in the U.S. or contact local emergency services now.

The honest answer to how many Americans will bet on the NFL in 2026 is that nobody knows, no public source we could find is counting it, and the figures standing in for it are measuring something else. That is less satisfying than 73.5 million. It is also the only answer the evidence supports.