What Are the Odds of Winning Solitaire? Klondike, Spider and FreeCell by the Numbers

"Is every game of solitaire winnable?" No. But the real answer depends a lot on which solitaire you mean. FreeCell is almost always winnable. Four-suit Spider often isn't. Klondike sits in between, and Draw 3 is noticeably tougher than Draw 1.

Rather than repeat numbers from around the web, we measured them. We pointed our own solver (the same one that picks the winnable deals and powers the Hint button on this site) at more than 40,000 deals and recorded what happened. Here is what it found, and how far you can trust it.

The short answer

Share of deals our solver won within its search limit. Klondike and Spider: deals numbered from 1 in our games. FreeCell: the classic Microsoft deals 1 to 32,000. "Proven lost" means the solver showed no winning line exists.

Read the "Won" column as a floor, not a ceiling. Every deal in it was actually won, and the winning moves were replayed through the game's rules to check them. The undecided deals are the solver saying "I ran out of time," not "this can't be done."

How we measured

A solitaire solver tries moves, backs up when it hits a dead end, and keeps going until it wins or runs out of options. Three details matter for understanding the numbers above.

The solver can see the face-down cards. In Klondike and Spider it knows every hidden card and the order of the stock. Researchers call this "thoughtful" solitaire. It answers the question "does a winning line exist?", which is a different question from "would a person find it?" FreeCell has no hidden cards, so there the two questions are the same.

It has a budget. Every deal gets a fixed number of positions to explore: 60,000 for Klondike, 100,000 for Spider and 200,000 for FreeCell. Most wins take a tiny fraction of that. If the budget runs out first, the deal is marked undecided.

Only some games can be proven impossible. The Klondike and FreeCell solvers consider every legal move, so when they run out of moves before the budget runs out, the deal really is unwinnable. The Spider solver skips moves that are almost never useful, to go faster. That makes it unable to prove a deal impossible, so Spider shows "n/a" in the "Proven lost" column.

Klondike: about 4 in 5, if you could see every card

Of the first 2,541 Draw 1 deals, the solver won 2,000 and proved only 28 impossible. The other 513 were undecided. So the true share of winnable Draw 1 deals is somewhere between 79% and 99%.

Draw 3 is harder. The solver won 2,000 of the first 2,819 deals (71%) and proved 246 impossible, nearly 9%. That is eight times as many dead deals as Draw 1.

Because Draw 1 and Draw 3 use the same layout for the same deal number here, we could compare them directly. Of deals 1 to 2,541:

• 1,698 were won at both Draw 1 and Draw 3.
• 302 were won only at Draw 1.
• 101 were won only at Draw 3.
• 440 were won at neither.

Those 101 deals are a useful honesty check. With unlimited passes through the stock, Draw 1 lets you reach any card that Draw 3 lets you reach, so in principle any deal winnable at Draw 3 is winnable at Draw 1 too. The solver simply ran out of budget on those 101 at Draw 1. That is the "undecided" bucket in action, and it is why we call these numbers a floor.

Our results line up with published work. Charlie Blake and Ian Gent of the University of St Andrews ran a much bigger search with their solver, Solvitaire, and put the share of winnable thoughtful deals at about 90% for Draw 1 and about 82% (81.9%) for Draw 3, the version they treat as standard Klondike. Both figures fall inside our ranges: 79% to 99% for Draw 1, and 71% to 91% for Draw 3. Their paper, The Winnability of Klondike Solitaire and Many Other Patience Games, was first posted in 2019 and published in the Journal of Artificial Intelligence Research in 2026.

Why you win less often than that

You don't get to see the face-down cards. Sooner or later every Klondike player faces a choice between two columns, picks one, and finds the card they needed under the other. A solver never makes that guess. Your real win rate depends on skill and on the version you play, but it will sit well below the thoughtful number. Don't measure yourself against 79%.

This is also why "Winnable deals only" is switched on by default here. Every new game comes from the list of deals the solver has won. It doesn't make the game easy, since you still can't see the hidden cards, but it means a loss was never decided by the shuffle.

Spider: the suit count changes everything

With one suit, the solver won 2,000 of 2,005 deals. Five went undecided, and we would bet most of those are winnable with more search. If you lose a one-suit game, it was almost certainly a misplay.

Two suits brings it down to 93% (2,000 of 2,150). Four suits is a different game: the solver won only 400 of 945 deals, about 42%, inside its budget. Remember that it can see every face-down card and the full stock order. That tells you how much harder four suits is for a person, and why winning a four-suit game is something to be proud of.

All three Spider numbers are lower limits, since the solver can't prove a Spider deal impossible. The real share of winnable four-suit deals is probably well above 42%. We just can't put a number on it yet. Our Spider strategy guide covers what changes as you add suits.

FreeCell: one bad deal in 32,000

FreeCell is the cleanest case because nothing is hidden. Windows numbered its FreeCell deals from 1 to 32,000, and we use exactly the same numbering. We ran all 32,000 through our solver:

• 31,996 were won within the 200,000-position budget, nearly all of them in well under a second.
• 3 (deals 6361, 8436 and 21554) needed a second run with a bigger budget, then fell too.
• 1, deal #11982, was proven impossible after the solver ran out of moves.

That matches what FreeCell fans have known for decades: #11982 is the only unwinnable deal in the original set. The whole run took about a quarter of an hour of computer time. The solver's typical winning line was about 52 moves long, though it doesn't look for the shortest line, and a good human solution is often tighter.

So when a FreeCell game beats you, the deal is almost never to blame. Hit Restart and try again. Our FreeCell strategy guide explains supermoves and walks through deal #1.

What the numbers mean for you

• FreeCell: treat every loss as solvable. It nearly always is.
• Spider 1 suit: the same. Undo is your friend.
• Klondike Draw 1: a good player wins plenty, but some losses come from hidden cards you couldn't have known about.
• Klondike Draw 3: expect more losses, and more deals that were never winnable at all.
• Spider 4 suits: the hardest of the lot. Winning a third of your games would be a good result.

The fine print

These figures describe deals in our games, under our rules: unlimited passes through the Klondike stock, and the standard Spider rule that every column must have a card before you deal. Other sites may use different rules or shuffles, and their numbers will differ. Search budgets are counted in positions, not seconds, so anyone running the same solver on the same deals gets the same results. We will update this page if we rerun the tests with larger budgets.