No. Some Klondike deals cannot be won however well you play. With every card visible, about 90% of Turn 1 deals and 82% of Turn 3 deals have a winning line; the rest are lost from the first move. Playing with the cards face down, as you normally do, you win fewer than that.
What "winnable" means
A deal is winnable if some sequence of legal moves gets all 52 cards onto the foundations. Whether it is depends only on the shuffle. To check it you need to know where every card is, including the 21 face-down tableau cards and the order of the stock. Researchers call this full-information version "thoughtful solitaire": the same rules, but with every card known from the start. A solver can play through millions of thoughtful deals and count how many have a solution.
That count is an upper bound for normal play. If a deal cannot be won with all cards known, it cannot be won blind either. The reverse does not hold: a deal can be winnable and still be lost because you had to guess.
How many deals are winnable
The most thorough study is by Charlie Blake and Ian Gent, whose program Solvitaire solved one million random Klondike deals with unlimited passes through the stock (The Winnability of Klondike Solitaire and Many Other Patience Games, Journal of Artificial Intelligence Research, 2026). They put Turn 3 winnability at 81.945% ± 0.084% and Turn 1 at 90.480% ± 0.116%. An earlier study by Ronald Bjarnason, Prasad Tadepalli and Alan Fern (Searching Solitaire in Real Time, ICGA Journal, 2007) had bracketed Turn 3 between 82% and 91.44%, so the newer figure sits at the bottom of that range.
Turn 1 is easier because every stock card is reachable on every pass. In Turn 3 only every third card lands on top, and a deal whose one useful card is buried in a group of three may have no way out.
Our own check, deal by deal
A percentage is nice to know, but it won't tell you whether the deal in front of you is one of the good ones. We wanted a list, so we ran a solver over this site's Klondike deals 1 to 100,000, once for Turn 1 and once for Turn 3, with all cards visible and a fixed search budget per deal. Every solution it found was replayed move by move through the game's own rules before the deal counted.
The result: 74,870 Turn 1 deals and 65,954 Turn 3 deals are proven winnable, and 59,874 are proven in both modes. These are lower bounds. A deal the solver gave up on within its budget is simply left out, so the true share is higher, and the research figures above say roughly by how much.
| Draw mode | Winnable with all cards visible (Blake & Gent, 1,000,000 deals) | Proven winnable on this site (deals 1 to 100,000) |
|---|---|---|
| Turn 1 | 90.480% ± 0.116% | 74,870 (74.9%), a lower bound |
| Turn 3 | 81.945% ± 0.084% | 65,954 (66.0%), a lower bound |
Those lists power the "Winnable deals only" switch on the Klondike page, under the Turn 1 and Turn 3 buttons. With it on, every new deal comes from the proven list for your draw mode. Deal numbers stay visible, so anyone with a solver can check our work. Deal 1, for instance, is on both lists.
Deals that are lost before you start
Some deals are dead on arrival. If no Ace is reachable, nothing in the tableau fits anywhere and the stock never offers a card you can use, there is no first move worth making. Others fail less visibly: a card sits on top of the very cards it would need to move onto, in a column nothing can clear. Bjarnason, Tadepalli and Fern ran a quick dead-end check over 10 million Turn 3 deals and caught 855,741 of them, 8.56%, as unwinnable before any search at all. The remaining unwinnable deals take a full search to rule out.
FreeCell is the opposite case. Every card is dealt face up, so "winnable" and "winnable for you" mean the same thing, and almost every deal qualifies: one count found 102,075 impossible deals among 8.6 billion, about one in 84,000. The famous exception is deal 11982, the one deal among Microsoft's original 32,000 that cannot be won (FreeCell on Wikipedia).
Winnable deals you still lose
All of the numbers above assume the player can see all 52 cards, and in a real game you can't. Twenty-one tableau cards start face down and the stock order is unknown, so some choices are guesses, and a guess can throw away a winnable deal. How often the best possible play wins under those conditions is an open question; Blake and Gent write that the answer "is not currently known for Klondike even within a factor of two". The firmest figure comes from Bjarnason, Fern and Tadepalli (ICAPS 2009), whose planning program, playing the normal game with hidden cards, won more than 35% of its deals. People tend to do worse than a good program. In a 2004 study by Yan, Diaconis, Rusmevichientong and Van Roy, an expert human won 36.6% of 2,000 games even with every card visible, about half of what the researchers' program managed.
That leaves a large group of deals that could be won and get lost anyway, usually to one early guess. That group is where better play pays off, and it is what the strategy guide is about.
FAQ
Is every Klondike solitaire game winnable?
No. With all cards visible, roughly 90% of Turn 1 deals and 82% of Turn 3 deals can be won. The others are impossible whatever you do.
How do I know if my deal is winnable?
You cannot tell from the layout; it takes a solver with full information. On this site, turn on "Winnable deals only" under the draw buttons and every new deal comes from a list of deals whose solution was found and replayed. Anywhere else, a lost game may or may not have been your fault.
Is Turn 1 or Turn 3 more often winnable?
Turn 1, by 8 to 9 percentage points when all cards are known: 90.5% against 81.9% in Blake and Gent's million-deal study. In real play the gap is wider, because hidden stock cards hurt more when you can only reach every third one.
Does unlimited undo make a deal winnable?
No. Which deals are winnable is fixed by the shuffle. Undo changes how many of them you win: each one lets you take back a guess once you have seen what it uncovered, so with enough patience you can, in principle, find the winning line of any winnable deal.