Klondike Solitaire

How to play Klondike Solitaire

Move all 52 cards onto the four foundations, building each suit up in order from ace to king. Clear every card and you win.

The deal lays out seven tableau columns of one to seven cards, with only the top card of each face up. The remaining 24 cards form the stock in the corner.

• On the tableau, build downward in alternating colors: a red six goes on a black seven.
• Tap or drag a card to move it; face-up runs move together. Turning over every face-down card is the real puzzle.
• Tap the stock to deal a new card to the waste. When the stock runs out, tap again to recycle the waste, as often as you like.
• Only a king can fill an empty column, and aces start the foundations.

Klondike solitaire strategy tips

• Flip face-down cards before anything else: a hidden card is worth more than a tidy foundation.
• Don't rush cards onto the foundations; you may need that red five to park a black four.
• Empty a column only when you have a king ready for it, ideally one with a useful run behind it.

What Klondike solitaire is

Klondike is the game most people mean when they say "solitaire." In Britain the same game is usually called patience, though that word covers the whole family of one-player card games rather than this one in particular. Klondike is simply the one people reach for by default, on both sides of the Atlantic. Seven tableau columns, four foundations, a stock you draw from in the corner: that's the shape most other games on this site borrow, so getting good at Klondike makes the rest easier to pick up.

Part of the reason it became the default is that Microsoft put it in the box. Solitaire shipped with Windows 3.0 in May 1990 and stayed in every release through Windows 7. Windows 8 dropped it in 2012, and Windows 10 brought it back in 2015 as the Microsoft Solitaire Collection. Twenty-two years in the box was enough to make "solitaire" mean this game and no other.

The short rules above cover what you're allowed to do. This is about what actually separates a win from a stuck board: the choices that matter, the mistakes that quietly end a game before you notice, and how often a deal is winnable in the first place. If you already know which draw setting you want, play turn 1 solitaire or turn 3 solitaire instead and skip the setting entirely.

Draw-1 vs draw-3 Klondike solitaire: which should you play?

The draw setting changes the character of the whole game. Draw-1 turns over one card from the stock at a time, so every card you haven't dealt into the tableau eventually becomes playable on its own. Draw-3 turns over three at once, and only the top card of each group can be played; the other two sit underneath it until you've cycled all the way back around to them. A card you need can end up buried two or three deep in a group, and getting it back to the top can cost you a full pass through the stock.

That difference isn't a matter of taste, and it has been measured. Give a solver x-ray vision so it can see every face-down card, and 90.480% of draw-1 deals turn out to be winnable against 81.945% of draw-3 deals. That is a gap of about 8.5 percentage points, produced by nothing but the draw rule. Draw-3 is the traditional tournament setting, and it is genuinely, measurably harder.

If you're new to Klondike, start with draw-1. If you want the version most competitions and most old desktop editions used, play draw-3. You can switch the setting on this page whenever you like, or go straight to turn 1 Klondike or turn 3 Klondike, which each lock the rule in and cover that setting in more depth.

Klondike strategy: flip first, foundation later

The biggest lever you control is the order you play in. Every tableau column starts with several cards face down, and until you turn them over you have no idea what's under there or whether it's the exact card you need next. Clearing a face-down card should almost always come before a tidy-looking move to the foundation, because information is worth more early on than progress. A move that flips a new card open keeps your options alive; a move that just rearranges a card you can already see rarely changes anything.

That same logic applies to the foundations themselves: don't send a card up just because it's eligible. A low card sitting on the tableau, a two or a three, is often more useful there than on the foundation, because it's the only thing that can catch a card one rank higher and one color different. Send it up too early and you can strand a card that had nowhere else to go.

Empty columns are valuable only if you fill them well. Only a king can start a new column, so don't empty one out unless you've got a king ready, ideally one dragging a useful run of cards with it rather than sitting alone. An empty column with nothing to put in it is a wasted slot, not an advantage.

Don't be shy about the stock, either. Once you've made every tableau move available to you, draw through it and recycle the waste as often as you need. There's no penalty for that, and undo is unlimited here, so you can back out of a bad recycle and try a different order.

Why Klondike solitaire games get stuck

Most lost games don't end with one dramatic mistake. They end with a string of small ones: a card sent to the foundation too early that turns out to be the only thing that could've caught a stray card later, a column emptied for a king with no run behind it, face-down cards left untouched too long while you chase easier moves until the board is packed tight with nowhere left to put anything. By the time it's obvious the game is stuck, the mistake that caused it usually happened five or ten moves earlier.

The fix isn't complicated: flip cards before you finish them, hold cards back if they might still be useful, and think about which column you're emptying before you empty it. If a line of play stalls out, undo a few moves and check whether an earlier one had a better option you skipped over.

How winnable is Klondike solitaire, really?

• 81.945%of draw-3 deals can be won - solver with perfect information Source
• 90.480%of draw-1 deals can be won - solver with perfect information Source

81.945% of draw-3 Klondike deals can be won, give or take 0.084%.

That figure is Charlie Blake and Ian P. Gent's, published in the Journal of Artificial Intelligence Research in February 2026. They built a solver for the game and ran it over 1,000,000 deals. 819,371 came out winnable, 180,472 unwinnable, and 157 the solver could not settle either way. You can read the paper in full.

Draw-1 got its own million-deal run and came out at 90.480%, give or take 0.116%: 904,226 winnable, 94,629 unwinnable, 1,145 undecided. Keep turning more cards at a time and the drop-off gets steep. Draw-2 is 88.620% winnable, draw-4 is 69.337%, and draw-5 is 53.434%, at which point nearly half the deals are lost before you touch them.

The rules the solver played under matter, because they are the rules you get here. A 24-card stock, drawn in groups, with redeals allowed without limit: recycle the waste as many times as you want. The authors are explicit that their solver cannot model a capped number of redeals at all, so if you play somewhere that gives you three passes and no more, none of these percentages describe your game.

Now the catch, and it is a big one. Every number above is measured with "thoughtful" play: the solver sees every face-down card before it starts. So it answers the question "could this deal be won?" and not the question "will you win it?" The paper says as much, and calls the second question open, noting that the true probability of winning from a starting position isn't known for Klondike even within a factor of two. Read 81.945% as a ceiling on real play, never as a forecast of your win rate.

The deals on this page are seeded, so the same seed always produces the same game and you can replay one exactly. They are random by default: we don't quietly filter out the unwinnable ones to keep your win rate looking healthy, which means some deals are lost the moment they're dealt, with nothing in the opening layout to warn you. If you'd rather not lose to the shuffle, turn on "Winnable deals only" in the settings menu and we'll deal you a game the solver has proved can be won. The hint button uses a solver too. When it can prove a move keeps a winnable game winnable, that's the move it shows you; when it can't, it gives you its best guess instead of pretending to know more than it does. If losing to the shuffle is what bothers you, FreeCell is the game to switch to: almost every FreeCell deal ever tested has turned out to be winnable.

How often do people actually win Klondike solitaire?

• 81.945%of draw-3 deals can be won - solver with perfect information Source
• 36.6%of 2,000 games won by expert human players shown every face-down card - 2004 study Source

Here is the most useful pair of numbers on this page, and the pair almost nobody puts side by side. About 82% of draw-3 deals are winnable in principle. In a 2004 study, expert human players who were also shown every face-down card won 36.6% of 2,000 games, give or take 2.78%. Same draw rule, same x-ray vision, less than half the result.

The gap is skill and search, not luck. Knowing where every card is doesn't tell you what order to play them in, and Klondike hides its mistakes well: the move that loses the game almost always looks reasonable at the time, and only turns out to be fatal ten moves later.

Computers playing the way you do, with the face-down cards genuinely face down, land in roughly the same place. In a 2009 study, one Monte-Carlo planner won 36.97% of 4,206 games (± 1.92%) and a second won 35.47% of 953 games (± 4.00%). A simple greedy bot in the same paper won 12.992% of a million games, and playing completely at random won 7.135%. Both papers are worth a look: the human study and the Monte-Carlo study.

Two different kinds of number just went past, and they are easy to confuse. The "can be won" figures - 81.945% and 90.480% - come from perfect information: the solver saw every card. The "did win" figures - 36.6%, 36.97%, 12.992%, 7.135% - come from playing games out and counting the wins. Sites that quote one and label it the other are why so many people think they ought to be winning four games in five.

Nobody has closed that gap. Persi Diaconis opened a 1999 colloquium talk at the University of Washington by saying: "One of the embarrassments of applied probability is that we can not analyze the original game of solitaire." More than a quarter of a century and several solvers later, that is still where things stand. We now know which deals can be won. We still don't know how often a person will win them.

How Klondike solitaire is scored

Klondike has no single official scoring system, which is why two apps can hand you different numbers for the same game. Standard scoring, the version most desktop editions inherited from each other, gives you points for moving a card to a foundation and for turning a face-down card over, takes points back if you pull a card off a foundation, and penalizes repeated trips through the stock. The exact values differ between versions, so a score is only meaningful against other scores from the same program.

Vegas scoring is the one worth understanding, because its arithmetic is fixed. You buy the deck for $52, a dollar a card, and earn $5 for every card you get onto a foundation. Ten cards up leaves you $2 down; eleven cards is your first dollar of profit; a complete 52-card win pays $260, a profit of $208. Because you pay for the deal itself, Vegas scoring punishes the habit of abandoning a game the moment it looks awkward, which is exactly how most people play.

This site doesn't score at all. It counts moves and time, which are the two things you can honestly compare between deals of the same game, and leaves the question of what a card on a foundation is worth to you.

Where Klondike solitaire came from

Nobody has documented why the game is called Klondike. The obvious guess is the Klondike gold rush of the late 1890s, and the timing is about right, but no one has produced any evidence for it. Treat it as a plausible story, not a fact.

The earliest known published rules are in Hoyle's Games (The McClure Company, New York, 1907), where the game appears on page 251 as "Seven-Card Klondike." The odd part sits three pages earlier in the same book: the game called plain Klondike, on pages 248 to 250, is a casino banking game dealt from a 13-card stock, which is what players now call Canfield, or Demon in Britain. At some point the two names traded places. You can read the 1907 book yourself.

Canfield is named after Richard A. Canfield, who ran the Saratoga Springs casino that now carries his name; the building was named for the man, not the other way round. He called his own game "Klondike." American rule books eventually settled on "Canfield" for the banking game and moved "Klondike" onto the seven-column game you are playing here. British usage went the other way, and "Canfield" there often means this game.

The reason everyone alive knows it, though, is Microsoft. Solitaire was written by Wes Cherry, a summer intern, in 1988. He never received royalties for it. The card faces were drawn by Susan Kare, who had done the original Macintosh icons. It shipped with Windows 3.0 in May 1990, and by 1994 the Washington Post was reporting that Microsoft officials said Solitaire had for years been the most-used Windows application. It stayed in the box through Windows 7, was dropped from Windows 8 in 2012, and returned in Windows 10 in 2015 as the Microsoft Solitaire Collection.

Klondike solitaire variants worth knowing

Change one rule and Klondike becomes a different game with a different difficulty. Double Klondike uses two decks shuffled together and eight foundations instead of four. Thumb and Pouch keeps the layout but lets you build a tableau card on any card of a different suit rather than only the opposite color, which loosens the game up considerably.

Whitehead deals every card face up, builds the tableau down in the same color, and lets you move any run as a block. Westcliff shortens the columns and starts you with the four aces already on the foundations. Easthaven swaps the draw rule out for Spider's: instead of a waste pile, the stock deals one new card onto every column at once.

Agnes is the confusing one. Two different games go by the name, with different rules, so check which one a site means before you decide its version is broken.

Klondike solitaire terms you'll run into

Foundation

The four piles at the top. You build up on them, ace to king, one suit each.

Tableau

The seven columns. You build down on them, alternating colors.

Stock

The face-down pile you draw from.

Waste

The face-up pile the stock feeds.

Talon

An older name for the stock - the same pile you're clicking in the corner.

Redeal

Also called a recycle: turning the waste back over to make a fresh stock.

"Build up" always means toward the king, and "build down" always means toward the ace, whichever pile is being described.

Talon turns up constantly in rule books and older software, and hardly ever in modern apps.

Some versions cap how many redeals you may do, at one pass or three. This one does not, which is worth knowing for a specific reason: unlimited redeals are exactly the rules the winnability study above measured. A three-pass limit is a genuinely different and harder game, and none of the published percentages apply to it.

Klondike solitaire FAQ