Play Microsoft FreeCell Online for Free

For an entire generation of computer users, FreeCell wasn't something you went looking for. It was just there — sitting in the Games folder of every Windows PC, quietly waiting between homework assignments and spreadsheet deadlines. No installation, no account, no tutorial. You clicked it, you figured it out, and before you knew it you'd burned an hour trying to clear deal #617.

The story starts in 1978, when medical student Paul Alfille created FreeCell on the PLATO educational computer system at the University of Illinois. He took an obscure older game called Baker's Game — which required building by suit and was brutally difficult — and made one elegant change: allow alternating-color stacking instead. That single rule tweak turned a niche puzzle into something almost anyone could learn to win.

FreeCell might have stayed an academic curiosity if not for Jim Horne, a programmer at Microsoft. In 1990, Horne wrote his own FreeCell implementation and convinced Microsoft to include it with the Windows Entertainment Pack for Windows 3.1 in 1991. Horne's version introduced the system of 32,000 numbered deals — a feature that would become central to FreeCell culture. Players could share deal numbers, compare strategies on the same layout, and systematically work through the entire set.

With Windows 95, FreeCell graduated from optional add-on to standard inclusion. It shipped with every copy of Windows 95, 98, ME, 2000, and XP — appearing on hundreds of millions of computers worldwide. The golden era of Windows FreeCell was the late '90s and early 2000s, when office workers, students, and retirees alike discovered the game through nothing more than idle curiosity and a Start menu.

Windows 7 (2009) was the last version to include the classic standalone FreeCell. When Windows 8 arrived in 2012, Microsoft replaced it — along with the other classic games — with the Microsoft Solitaire Collection, a free-to-play app bundling FreeCell, Klondike, Spider, Pyramid, and TriPeaks. The collection works fine, but it introduced something the original never had: advertisements. Video ads play between games, banner ads line the interface, and removing them requires a monthly subscription. For many longtime players, it just wasn't the same.

That's what this site is for. The same clean, distraction-free FreeCell you remember — running in any modern browser, on any device, with no ads interrupting your gameplay.

This isn't just “a FreeCell game.” We use the same pseudorandom number generator and dealing algorithm that Jim Horne implemented in the original Microsoft FreeCell. That means games #1 through #32,000 on this site produce the exact same card layouts as the Windows version.

Had a favorite deal? A nemesis deal? A number you remember beating on a rainy afternoon in 2003? Type it in and you'll get the same cards in the same positions. Deal #1 is the same gentle opener. Deal #11982 is the same impossible puzzle. Every number in between is a pixel-perfect recreation of the original layout.

You can browse all available deals in the Deal Explorer, which lets you search by number, filter by difficulty, and jump straight into any game. It's the “Select Game” dialog from Windows FreeCell, but better.

Of the original 32,000 Microsoft FreeCell deals, exactly one has been proven impossible to solve: Deal #11982.

The discovery is one of the great collaborative stories of early internet culture. In 1994, Dave Ring launched the Internet FreeCell Project, coordinating thousands of volunteers who systematically worked through all 32,000 deals. By 2000, every deal had been solved at least once — except #11982. Exhaustive computer analysis later confirmed what the volunteers suspected: no sequence of legal moves can clear the board. Every path leads to a dead end.

Eight other deals (#146, #455, #495, #512, #530, #1941, #6182, #8591) were long considered impossible but were eventually cracked using advanced solvers. Deal #11982 remains the lone holdout. You can try it yourself — just don't expect to win.

Most solitaire games involve a fair amount of luck. You flip cards from a stock pile and hope for the best. FreeCell is fundamentally different: all 52 cards are dealt face-up from the start. There is no hidden information, no luck of the draw, no blaming the shuffle. When you lose, it's because of your decisions — and when you win, you earned it.

The four free cells — temporary parking spaces for individual cards — give you just enough room to maneuver without making the game trivial. Expert players learn to keep cells empty as long as possible, using them as a last resort rather than a first move. The tension between needing free cells to execute a plan and needing them empty for future flexibility is what gives FreeCell its strategic depth.

Then there were the numbered deals. That “Select Game” dialog — a simple text field where you typed a number from 1 to 32,000 — turned FreeCell from a random pastime into a shared experience. You could tell a coworker “try deal #1941” and know they'd face the exact same challenge. People kept notebooks of beaten deals. Online forums traded strategies for specific numbers. Speed records were set and broken on particular deals. No other solitaire game created that kind of community around individual puzzles.

And it was always there. You didn't have to install anything, buy anything, or sign up for anything. Click Start, click Games, click FreeCell. That frictionless accessibility — combined with a game that rewarded genuine skill — is why Microsoft FreeCell became one of the most-played computer games in history.

We built this site because we missed the original too. Here's how the two versions compare:

The thirty-two thousand numbered deals that defined Microsoft FreeCell are not arbitrary. They are the output of a specific algorithm: a seeded linear-congruential pseudorandom number generator, an integer modulus of 2^31 − 1, and a specific deal-out procedure that walks the shuffled deck across the eight tableau columns one card at a time. Feed the algorithm a seed — any integer from one to 32,000 — and you get a deterministic, reproducible layout. Deal #1 is always the same board. Deal #617 is always the same board. The set is frozen by the algorithm, not by a stored database.

Why cap the set at 32,000? The original Windows FreeCell game dialog only accepted values up to 32,000 because 2^15 − 1 (which equals 32,767) is the maximum positive value of a signed 16-bit integer, and the rounded-down thirty-two-thousand ceiling fit neatly inside that range. That is the entire explanation. There is nothing mathematically special about 32,000. It was a pragmatic bound that matched the data type the dialog box used.

The PRNG itself was the classic Microsoft C-runtime generator: seed × 214013 + 2531011, then shift right by sixteen bits and mask with 0x7FFF. Horne used that generator to produce a sequence of pseudorandom indices into the remaining deck, and the sequence was used to deal cards one at a time across the eight columns. Because the generator is deterministic and widely documented, modern implementations (including this one) can reproduce the original Microsoft deals exactly. That is why deal numbers from Windows FreeCell still work here — we run the same arithmetic.

The deterministic property of the set created something Horne did not foresee: a shared cultural vocabulary. When two FreeCell players from different continents both said “I could not beat #617,” they were talking about the same board. The deal number became the stable identifier for a puzzle, the way chess positions have FEN strings or crosswords have puzzle numbers. Almost no other solitaire game of the era had that property, and it is the single most important reason FreeCell developed the collective lore it did.

One useful consequence of the shared numbering system: it means a beginner and a tournament player can literally play the same board. Chess has that property — a grandmaster and a novice can both study the same position — and so do crosswords, but very few casual card games do. FreeCell's numbered deal set turned every deal into a potential teaching example, a potential tournament position, and a potential community anecdote. That is why deal numbers survive as cultural currency more than three decades after the original ship date.

The extension beyond 32,000 came later. When modern implementations (including ours) wanted to go past Horne's original ceiling, the natural choice was to keep the same PRNG and dealing algorithm but extend the seed range up to 2^31 − 1, which gives more than two billion distinct deals. The Microsoft 32,000 remains the canonical set — it is the range with shared history — but anyone looking for fresh puzzles can draw from the much larger extended range without leaving the Horne-era algorithm behind.

Once the deal-number system existed, specific deals became famous. Some earned reputations for difficulty, some for elegance, some for being beginner traps or welcome gifts. Here are the deal numbers that FreeCell players mention most often, with brief notes on why each one matters and how to think about it.

Deal #11982 — the unsolvable one

The most famous individual deal in FreeCell history. It is the only deal in the original Microsoft 32,000 that has been proven genuinely unwinnable. Every legal sequence of moves leads to a dead-end position, and no amount of creative play can salvage it. Read the full walkthrough on our deal #11982 page, or try to bang your head against it at /game/11982.

Deal #617 — the folklore nemesis

A deal with a long reputation as a difficult but solvable puzzle. Community archives from the 1990s are full of posts about #617, because it sits at the boundary where human search begins to struggle but the solution is still within reach. Players who beat #617 remembered it. The deal is technically solvable and the solution was verified decades ago, but finding the line without solver help is a genuine test.

Deals #1 through #3 — the welcome mat

The first three deals in the set are unusually gentle, which is not accidental: Horne chose the seed scheme so that low-numbered deals tend to be forgiving. Deal #1 is a nearly-trivial board that most players finish in under two minutes. Deals #2 and #3 are slightly harder but still within reach of a first-time player. The welcome-mat effect was deliberate: the first deals you ever saw were designed to let you win and come back.

Deal #6974 — a tournament favourite

A deal often chosen for speedrun and tournament events because it is challenging without being brutal. The opening is interesting, the midgame has multiple viable lines, and the endgame rewards counting. Tournament organisers like deals like #6974 because they produce meaningful differentiation between strong and average players without the random-feeling difficulty of the genuinely hard deals.

Deals #146, #1941, #6182, #8591 — the reclassified

These deals were long considered impossible by the Internet FreeCell Project before modern solvers eventually proved them solvable. Their winning lines are counterintuitive — sacrificial cell usage that human players reject on sight. Studying their solutions is a lesson in how often intuition about solvability is wrong.

For a longer list of notable deals, visit our famous FreeCell game numbers page.

Studying named deals is one of the most efficient ways to improve. A player who works through the standard list of famous deals will encounter almost every opening pattern, every mid-game cell trap, and every endgame ordering puzzle that FreeCell can produce. Good players often keep a list of deals they have beaten, deals they have lost, and deals they intend to revisit, and they use that list the way chess students use their own game notebooks.

Jim Horne's port of FreeCell arrived in 1991 bundled with the Windows Entertainment Pack Volume 2 — an optional add-on for Windows 3.x — and it stayed roughly unchanged for more than two decades. The port took Paul Alfille's PLATO-era design, added the 32,000 numbered deal system, and put it in front of an audience measured in hundreds of millions. By Windows 95, FreeCell was a standard install: if you had a Windows PC, you had FreeCell, whether you wanted it or not.

Workplace productivity research from the 1990s and early 2000s repeatedly listed FreeCell and Minesweeper as the most-played computer games in the world by raw play count, not because anyone had branded them that way but because they were on every desk. A generation of office workers learned FreeCell on their lunch breaks. A generation of students learned it in computer labs. Cultural footprint at that scale is rare for a deterministic card puzzle.

Microsoft kept FreeCell in Windows 98, ME, 2000, XP, Vista, and 7, all with substantially the same implementation. The only material updates across that twenty-year run were cosmetic — slightly redrawn card backs, a clearer dialog, sharper fonts. The gameplay was stable enough that players who took a decade off could come back and find their favourite deals exactly as they had left them.

Windows 8 (2012) ended the classic run. Microsoft replaced the standalone game with the Microsoft Solitaire Collection, which bundled FreeCell alongside Klondike, Spider, Pyramid, and TriPeaks, and which introduced advertising into a previously ad-free experience. The Solitaire Collection still supports the original 32,000 deals, but the experience around them changed materially: video ads between games, banner ads during play, and a subscription model to remove both. For many longtime players, the Windows 7 era is the endpoint of “classic” Microsoft FreeCell.

The shift was controversial at the time and remains a touchstone in community discussions. Players who had grown up with the clean, ad-free version described the change as the end of a small but real part of the Windows identity. Microsoft had good reasons — bundling games together reduced maintenance burden, and the free-to-play model funded ongoing updates — but the shift illustrated how quickly a default install becomes infrastructure. FreeCell had been part of the Windows PC the way the taskbar was, and making it feel commercial altered the texture of the whole operating system for some long-time users.

Most games in the original Windows Entertainment Pack are gone. Minesweeper lives on in corners of the web and in reissues, but Taipei, JezzBall, Rodent's Revenge, Tetris for Windows, SkiFree, and a dozen others are effectively dead. FreeCell is the rare survivor. Why?

Three design strengths explain it. First, full visibility: FreeCell is a puzzle, not a gamble, and that means every game rewards thought in a way that players come back to. Solitaire games with hidden information feel random when they go badly; FreeCell never does. Second, the numbered deal set: shared identifiers create shared stories, and shared stories create community. Third, the near-perfect solvability rate: FreeCell is one of the few card games where you can reasonably expect to win ninety percent of games if you play carefully, which keeps the satisfaction feedback loop tight.

Those three design strengths were not accidental. Paul Alfille explicitly set out to design a solitaire variant that rewarded planning over luck, and Jim Horne preserved that design intent while adding the deal-number infrastructure that let players share their experience. Good design ages well, and FreeCell is the clearest case we know of a card game whose design intent still drives its popularity thirty-plus years later.

There is a fourth reason worth naming: FreeCell scales beautifully. New players finish the early deals and feel competent. Intermediate players get stuck on specific deals and have something concrete to return to. Strong players chase the solver ceiling and track their own win rates. Tournament players compete on hardest-deal sets. Every skill level has a natural next challenge built into the same game, and the numbered deal set makes those challenges legible. That is a kind of design longevity that most puzzle games never achieve.

The Entertainment Pack games that disappeared tended to lack at least one of those features. Minesweeper had a similar skill ceiling but no shared identifiers, so community collapsed when the game moved off the default install. JezzBall had no long-form depth. Taipei had no numbered boards. FreeCell had all four: design intent that rewarded skill, numbered deals, near-perfect solvability, and a difficulty curve that scales with the player. That combination is why it survived.

FreeCell is the centre of a small family of closely related games. Each variant changes one rule and produces a noticeably different experience. Understanding the family is part of understanding FreeCell itself, because the tradeoffs Alfille made are clearer when you see what happens when you undo them.

Baker's Game — the ancestor

Before FreeCell, there was Baker's Game: the same eight-column layout, the same four cells, but with same-suit tableau stacking instead of alternating-colour. Baker's Game is significantly harder because same-suit stacking removes half of your legal receiving cards. Solvability drops from 99.99 percent to around seventy-five percent. Alfille's single change — to alternating-colour stacking — is what turned a niche hard puzzle into a mass-audience game. See the full Baker's Game rules for the details.

Eight Off — extra cells, stricter stacking

Eight Off doubles the number of cells to eight and uses same-suit stacking. The extra cells more than compensate for the stricter stacking, producing a solvability rate around eighty-nine percent. It is a slower, more methodical game than FreeCell, because eight cells makes aggressive planning almost unnecessary. Read more on our Eight Off page.

Seahaven Towers — ten columns, four cells

Seahaven uses ten columns of five cards each, four cells, and same-suit stacking. The narrower columns and same-suit rule make it harder than standard FreeCell, but the extra columns provide compensating structure. See our Seahaven page for the full rules.

Penguin — different tableau, same cells

Penguin is a FreeCell relative that keeps the four-cell structure but starts with a tableau of seven columns of seven cards plus a distinctive “flipper” row. The Aces begin on foundations, which changes early strategy substantially: you are not hunting for Aces, you are building tableau sequences that eventually feed the already-started foundations. Penguin is moderately harder than FreeCell and has a loyal niche following.

Cell-variant FreeCell

Shrinking the cells from four to three, two, or one produces a family of difficulty variants. Three-cell is marginally harder. Two-cell is noticeably harder, with solvability around eighty-five percent. One-cell is brutal, with solvability around ten percent. These variants are covered in detail on the FreeCell variants page, which also discusses Penguin, a tableau variant with a different starting configuration. If you want to chase speed and streak records across any of these, see our FreeCell world records page.