FreeCell: almost every deal can be won — so what's stopping you?

FreeCell is the friendliest kind of hard. Every card is face up from the first second, nothing is hidden in a stock pile, and almost every deal has a winning path. When you lose a game of FreeCell, it's rarely the cards' fault.

We checked that claim ourselves. Our solver played 10,000 shuffled FreeCell deals under this site's rules and found a win in 99.68% of them. It didn't prove a single deal impossible; the remaining 0.32% simply weren't finished within the search limit we set.

So for a casual player the real question isn't "can this be won?" It's "what habits make me lose a game that could be won?" That's what this page is about.

What our FreeCell tests found

The fewer-cells rows are there for a reason. A cell with a card parked in it is, for the moment, a cell you don't have. Fill two cells carelessly and you're playing a game where only about three deals in four can still be won. Fill three and you're down to fewer than one in five.

How FreeCell is played here

• Eight columns hold all 52 cards, face up: four columns of seven and four of six.
• Four free cells, top left. Each holds one card of any kind.
• Four foundations, top right, built up by suit from Ace to King.
• In the columns, build down in alternating colors: red Jack on black Queen, black 3 on red 4.
• Empty columns accept any card.
• Moving runs. You can move several cards at once when you have enough empty cells and columns to shuffle them through one at a time. The game works this out for you.

Win by getting all four suits onto the foundations.

Playing on instinct vs. playing to the solver's standard

We didn't build a casual FreeCell bot, and there's a good reason. In a game where nearly every deal is winnable, the interesting number isn't how often a naive player wins — it's how quickly that player throws away the space they need. The fewer-cells tests are our stand-in for that. They show, precisely, what each wasted cell costs.

The drop from three cells to two is where things fall apart. That's the practical rule: try never to have more than one cell occupied for long.

Five habits of players who rarely lose

Take a full look before the first move

Only about half of deals (49%) let you put an Ace up straight away. Whether yours does or not, find all four Aces and count how many cards sit on top of each. The most deeply buried Ace is usually where your plan starts.

Treat free cells like borrowed money

Parking a card is easy. Getting it back out needs a home for it. Before you use a cell, know where that card is going next.

Go after an empty column

An empty column is worth more than a cell. It holds a whole run, not a single card, and on this site each empty column adds one to how many cards you can move at once (the limit is empty free cells + empty columns + 1). The six-card columns are the easiest to clear.

Don't rush low cards home

A red 4 on the foundation can't hold a black 3 for you anymore. If you still have cards of the opposite color two ranks down buried somewhere, wait.

Undo is your safety net

With no hidden cards, every position can be worked out — so if you hit a wall, undo back to the last real decision and take the other branch. Given how few deals are truly impossible, it's almost always worth it.

The mistakes that cost winnable games

• Filling three or four cells "just for now."
• Putting a random card into an empty column with no plan for it.
• Burying a low card under a long run you can't easily lift later.
• Restarting instead of backing up a few moves and thinking.

FreeCell on this site

Free to play in any modern browser, on a computer, tablet or phone. You don't download anything, create an account or hand over an email address. The cards are large and clear, there's zoom for bigger cards, and you have Hint, unlimited Undo, Restart and Auto Play. Drag and drop, or tap a card then tap its destination. Wins, time and moves are tracked for the current game; sound can be switched off; fullscreen and table colors are in the settings. There's no gambling and nothing to buy.

Want a harder variant with the same layout? Baker's Solitaire builds by suit and is much tougher on this site. Prefer a bit of luck? Go back to classic Klondike, or take on Yukon.

The research behind these numbers, step by step

The play-solitaire-free-online.us team at Solitairen OOD ran these tests in September 2026 on FreeCell exactly as this site plays it.

1. Rules from our game code. Eight columns (7,7,7,7,6,6,6,6), four cells, alternating-color builds, any card into an empty column, multi-card moves limited by free space. That move limit can't change which deals are winnable, because the same move can always be made one card at a time.
2. Game rebuilt in code. We wrote FreeCell as a program the solver could search through.
3. Numbered deals. Seeds 1 to 10,000 for the main run; 3,000 deals each for the 0-, 1-, 2- and 3-cell tests. They aren't the old Windows deal numbers.
4. The player. A solver that always looks at the most promising position next, with a limit of 200,000 positions per deal. It only makes automatic foundation moves that are provably safe.
5. A second pass. The 87 deals still unsolved after the first run were tried again with a limit of 1,000,000 positions. That brought the solved share to 99.68%, with 32 deals left undecided and none proven impossible.
6. Published comparison. Blake & Gent (2026) give 99.998881% for 4 cells, 99.36% for 3, 79.54% for 2, 19.35% for 1 and 0.21% for 0. Each of our ranges contains the published value.
7. Margin of error. The "95% range" is the band the true figure almost certainly falls in; for 4 cells, 99.55% to 99.77% for what our solver finds.
8. Cross-check and limits. We re-tested a handful of "impossible" verdicts from the fewer-cell runs with a completely separate brute-force program (FreeCell 1-cell deals 1 to 5, and 3-cell deals 739 and 1053); it agreed on all of them. One bias remains: this site lets a card come back off a foundation, and our solver never does that, so our figures can only err on the low side.

The code behind the test

Everything runs in Python 3.11, standard library only. FreeCell (and Baker's Solitaire) live in freecell.py. The solver is a best-first search: it scores each position by how many cards are still off the foundations and how many sit on a lower card, and always expands the best-scoring one next. A table of positions already seen stops it going round in circles. runner.py shares the 10,000 deals across 19 worker processes on a 20-core machine, and the main 4-cell run took about 18 minutes; the bigger second pass was handled by retry.py. Percentages come with Wilson score intervals.

We had help from an AI assistant, Claude by Anthropic, in writing and reviewing the solver and in drafting this page. None of the figures came from the AI: they came from running the solver, and we checked them against the published study.

By Admin, play-solitaire-free-online.us (Solitairen OOD). Research dated September 18, 2026.

Questions new players ask

If nearly every deal is winnable, why do I lose so often?

Most often, the cells got clogged. Our tests show that playing with two cells tied up is like playing a game where about a quarter of deals can't be won at all.

Is there any FreeCell deal that truly can't be won?

A tiny number. Our solver didn't prove any of its 10,000 deals impossible; the published estimate is that about 1 in 90,000 deals can't be won.

How many moves does a game take?

Our solver's first solutions averaged about 62 moves. They weren't the shortest possible, and people usually take more, because we try things and undo them.

Should I use Hint?

It's there to help and there's no penalty. Many players use it only when they're certain they've missed something.

What's the best game to learn after FreeCell?

Spider Solitaire uses the same idea of building long runs, with two decks and a stock to deal.