Ranking the most popular video games by copies sold answers a question nobody is really asking. A game that sold eighty million copies a decade ago and is now played by nobody outranks a game that millions of people open every evening, because one of those numbers accumulates forever and the other does not.
Ranking by players is the more useful question and the harder one, because almost nobody publishes player counts. Two sources do. Valve publishes a live chart of concurrent players for every game on Steam, and Roblox reports average daily active users to its shareholders because it is legally obliged to. Everything else, including the three games most people would name first, is a press release or an estimate.
Most Popular Video Games on Steam by Peak Concurrent Players
These are peak concurrent players for a single day, 6 October 2026, from Valve’s own chart. A concurrent count is how many people were in the game at the same moment, so it is a far smaller number than how many people played that day.
| Rank | Game | Peak players in one day | Model | Age |
|---|---|---|---|---|
| 1 | Counter-Strike 2 | 1,140,811 | Free to play | Series dates to 1999 |
| 2 | Dota 2 | 826,202 | Free to play | Released 2013 |
| 3 | PUBG: Battlegrounds | 612,531 | Free to play | Released 2017 |
| 4 | AION 2 | 397,905 | Free to play | Recent release |
| 5 | Apex Legends | 196,754 | Free to play | Released 2019 |
| 6 | WARDOGS | 194,836 | Paid | Recent release |
| 7 | FiveM | 183,928 | Free to play | GTA multiplayer modification |
| 8 | Bongo Cat | 178,886 | Free to play | A desktop toy, not a game |
| 9 | Deadlock | 155,779 | Free to play | In development |
| 10 | Delta Force | 125,815 | Free to play | Recent release |
What the Steam Chart Measures and What It Misses
The Steam Most Played chart is the closest thing the industry has to an audited popularity ranking, because Valve counts sessions on its own platform and publishes the result in public, continuously. It is a concurrent user count, which means it shows how many people were playing simultaneously rather than how many played that day, and the two differ by a large multiple.
Two things about the list above are more revealing than the ranking itself. The first is that nine of the ten are free to play, which tells you that on Steam the popularity ceiling belongs to games with no purchase in the way. The second is that one entry, sitting eighth, is not a game at all. Bongo Cat is a desktop toy that runs as a Steam application, and it holds a position that would otherwise belong to a major release.
The bigger limitation is coverage. The chart only knows about Steam, so it says nothing about console players, nothing about mobile, and nothing about the launchers Steam does not run. A game can be enormous and invisible here, which is exactly what happens with the three titles most people would put at the top of any popularity list. Our guide to the best selling consoles of all time runs into the same problem from the hardware side.
Roblox Is the Only Other Game That Publishes Its Numbers
Roblox reports daily active users every quarter because it is a listed company and the metric is material to its business. In its Form 10-Q for the quarter ended 30 June 2026, the company reported an average of 123 million daily active users, of whom roughly 1.6 million were paying on an average day.
That figure is not comparable with the Steam chart and it is worth being clear about why. Daily active users counts everyone who opened the platform across a whole day on any device, while a concurrent count measures one instant. Putting 123 million next to 1.1 million as though they rank against each other is the single most common error in articles on this subject.
What makes the Roblox number unusually trustworthy is that the company documents how it is built. The same filing explains that Roblox continually refines its methodology to exclude botting, non-active time and unauthorised use, and warns that these refinements change the reported figures. A company that publishes the weaknesses of its own metric is giving you more than a press release ever does.
There is a second pattern in the top ten that is easy to miss. Four of the ten are either recent releases or still in development, and they sit alongside games that have been running for a decade or more. A popularity chart measured in concurrent players is therefore reading two different things at once: the enduring audience of a long-lived game, and the opening weeks of a new one, which can look identical on a single day and look nothing alike a year later.
The Games Everyone Names and Nobody Can Verify
Ask anyone for the most played games in the world and you will hear Minecraft, Fortnite and Grand Theft Auto. None of the three has a continuously published player count. Epic Games is private and releases Fortnite figures when it suits a announcement. Minecraft belongs to Microsoft, which reports Xbox as revenue rather than players. Take-Two reports units shipped for Grand Theft Auto, not people playing it.
- Steam concurrents are public, continuous and audited by the platform that counts them.
- Roblox daily active users are public because securities law requires it.
- Fortnite, Minecraft and Call of Duty numbers appear only in announcements.
- Mobile figures usually come from analytics firms estimating from the outside.
- A number with no date and no publisher attached is an estimate wearing a suit.
- Concurrent counts and daily counts are different units and cannot be ranked together.
This is why two popularity lists almost never agree. One builds on Steam data, another on download estimates, a third on a three-year-old press release, and each presents the result as a ranking. The honest version says which numbers are real. Our guides to the Nintendo Switch 2 games and the best AAA games on iPhone cover the platforms this chart cannot see.
How to Read a Popularity Ranking
The practical test is short. Ask what was counted, when, by whom, and over what period. A concurrent peak on a named date from the platform itself is evidence. A round number with no date is marketing. Anything described as players worldwide without a source is usually an installed-base estimate, which counts people who downloaded a game once and never returned.
One more habit helps. Check whether the figure is a peak or an average, because a peak is the single best moment of a day and an average is the whole of it, and the two can differ by half. The Steam chart publishes both a current figure and a daily peak side by side, which is why the table above says peak in the column heading rather than leaving you to guess.
It also pays to notice what a chart structurally cannot show. Steam will never surface a console exclusive, and Roblox reports a platform rather than a game, so its 123 million covers millions of separate experiences made by its own users. Both numbers are real. Neither answers the question on its own, and the gap between them is where most of the confusion in this subject lives.
Frequently Asked Questions
What is the most played game on Steam?
Counter-Strike 2, by a wide margin, with a daily peak above 1.1 million concurrent players on 6 October 2026. Dota 2 is second at around 826,000 and PUBG third at around 612,000. All three are free to play and all three are old, which is the pattern on Steam: the top of the chart belongs to long-running free-to-play games rather than new releases.
Is Roblox bigger than the games on the Steam chart?
Almost certainly, but the numbers do not prove it directly because they measure different things. Roblox reported 123 million average daily active users for the quarter ended 30 June 2026, which counts everyone who opened it on any device during a day. Steam publishes concurrent players, a snapshot of one moment. Comparing the two as if they were the same unit is wrong.
Why is Minecraft or Fortnite not on a player-count list?
Because neither publishes continuous player counts. Epic Games is privately held and shares Fortnite figures only in announcements, and Microsoft reports Xbox performance as revenue rather than player numbers. Both games are certainly played by enormous numbers of people; there is simply no published, dated figure to put in a table beside the ones that exist.
Do concurrent players tell you how many people play a game?
No, and the difference is large. A concurrent count is how many people were playing at the same instant, so a game peaking at a million concurrent players is being played by several million people across a day, depending on how long sessions run and how spread out they are across time zones. Concurrents are useful for comparing games measured the same way, not for estimating an audience.
