How To Play FreeCell
Setup
All 52 cards are dealt face up into 8 columns (the first four get 7 cards, the rest 6).
Above are four free cells — each holds one card as temporary storage — and four foundations.
Rules
Build the tableau down in alternating color (a red six on a black seven); build each foundation up by suit from the Ace. Any card fits an empty column.
You move one card at a time, but a valid alternating-color run can be moved as a group up to the supermove limit — (1 + free cells) × 2^(empty columns) — which is just what you could do one card at a time through your free cells and empty columns.
Clear all 52 to the foundations to win.
Difficulty is the number of free cells: 4 (standard) or 2 (hard).
Odds
The solitaire you're almost always supposed to win — which is exactly what makes a loss sting. A look at the numbers.
The shape of a deal
All 52 cards are dealt face-up into eight columns — four of seven cards and four of six — with four free cells and four empty foundations. Nothing is hidden. There is no stock to draw, no card you haven't seen. Whatever happens from here is information you already had.
Almost every deal is winnable
This is FreeCell's famous quirk, and the reason it became the solitaire of choice for a generation: with four free cells, very nearly every deal can be solved. It is one of the most-studied card games in computing, and the verdict is overwhelming.
Of the 32,000 classic numbered deals, exactly one was ever proven unwinnable (the legendary game #11982). Our own solver tells the same story from the other side: given a little room to think it finds a win on 9 deals in 10, and the more it's allowed to search, the higher that climbs — it never bumps into a ceiling short of "solved."
| Solver effort | Wins found (4 cells) |
|---|---|
| a little | ~80% |
| more | ~91% |
| …and rising | → ~100% |
So it really is all skill
If nearly every deal can be won, then a loss isn't the cards — it's the line of play. To make the point concrete: a bot that only ever makes the obvious forward move, never planning a sequence, essentially never finishes.
FreeCell gives you every card and every option in plain sight, then asks the hardest question a solitaire can: not what can you see, but how far ahead.
Two cells is a different game
The game's one dial is the number of free cells — four or two — and it is not a gentle nudge. Dropping to two cells doesn't just make deals harder; it makes many of them impossible. Handed the same effort, our solver's success rate roughly halved at two cells, and a large share of deals that are almost trivial with four simply have no solution with two.
Why cells matter so much: a free cell isn't just a parking spot —
free cells and empty columns are the machinery that lets you shift a whole run at once.
The most you can move as a group is (1 + free cells) × 2empty columns.
That's why an empty column is worth more than an empty cell — it doubles your
reach.
| Free cells | Empty cols | Cards you can move at once |
|---|---|---|
| 4 | 0 | 5 |
| 4 | 1 | 10 |
| 2 | 0 | 3 |
| 2 | 1 | 6 |
How to beat the odds
- Keep cells empty. Every occupied cell shrinks how much you can move — a full cell is borrowed space you have to pay back.
- Fight for an empty column. It's the single most powerful thing on the board; it doubles your moving power and can hold a King to start a fresh run.
- Chase the low cards. The foundations climb from the Aces up, so buried Aces and twos are what stall a game. Free them before you tidy the high end.
- Look before you bank. A card sent home can't come back — and just because the game can be won doesn't mean this move keeps it winnable.
The ~99.99% figure is the long-established result for standard four-cell FreeCell across exhaustive study of millions of deals; #11982 refers to the classic Microsoft numbering, not this game's shuffles. The "wins found" and "0 / 3000" figures come from our own solver and a forward-only bot run over this game's engine; the solver is a lower bound — it finds real wins but gives up under a search budget, so the true solvable rate sits above what it reports and rises the longer it looks.
