Short answer: the cards do not change, but your access to them does. We dealt the same 10,000 numbered Klondike deals twice and let a simple bot play them — a bot that sees only face-up cards and never takes a move back. At Turn 1 it won 42.3% of those deals (95% range 41.3–43.3%). At Turn 3, on the very same deals, it won 10.6% (95% range 10.0–11.2%). Four times fewer wins, same shuffle.
That single comparison is the reason this site is called Turn 3. Below we show what we measured, what published research says, and how to play the harder rule without feeling punished by it.
The headline numbers
We ran two different kinds of test. One is a floor, one is a ceiling, and the truth for a human sits between them.
1. The blind bot: a floor, not how often people win
Our first test used 10,000 numbered deals and a deliberately simple player: it sees only the face-up cards, follows plain priority rules, and never undoes a move. Every deal was played once as Turn 1 and once as Turn 3.
| Draw rule | Deals | Bot wins | 95% range |
|---|---|---|---|
| Klondike Turn 1 | 10,000 | 42.3% | 41.3–43.3% |
| Klondike Turn 3 | 10,000 | 10.6% | 10.0–11.2% |
Treat those as a floor. A bot with no memory, no planning and no Undo is a much weaker player than you are. It tells you how much harder the rule makes the game for someone playing blind — it does not tell you your own win rate.
2. The solver: a ceiling, and an honest range
Our second test asked a different question: could the deal be won by someone who knew where every card was? We gave a solver 1,000 deals and a budget of 150,000 positions per deal.
| Result on 1,000 deals | Turn 1 | Turn 3 |
|---|---|---|
| Solved (a win was found) | 76.6% (73.9–79.1%) | 66.2% (63.2–69.1%) |
| Proven impossible | 2.1% (1.4–3.2%) | 9.5% (7.8–11.5%) |
| Undecided within the budget | 21.3% (18.9–24.0%) | 24.3% (21.7–27.1%) |
Two things jump out. First, proven-impossible deals more than quadruple under Turn 3, from 2.1% (1.4–3.2%) to 9.5% (7.8–11.5%). Some deals are simply dead when you can only reach every third stock card. Second, a chunk of deals stayed undecided — the solver ran out of budget before proving anything. That means the winnable share is a range, not a number: for Turn 3 it lies somewhere between the solved 66.2% and 66.2% plus the undecided 24.3%.
How we did it: our team ran these simulations in September 2026 against this site's own Klondike rules, in Python, with an AI assistant (Claude, by Anthropic) helping write the code and the drafts. Every figure above comes from running that code — nothing is estimated.
What the published research says
Our solver ceiling lines up with the best academic work on the question. In February 2026, Charlie Blake and Ian Gent published The Winnability of Klondike Solitaire and Many Other Patience Games in the Journal of Artificial Intelligence Research. Using far more computing power than we did, they put winnability at 90.48% for one-card draw and 81.95% for three-card draw, in the "thoughtful" version of the game where the player already knows the rank and suit of every card. Their Klondike figure is reported as 81.945% ± 0.084%.
Now put the two views side by side. If you can see through the cards, three-card draw costs you roughly 8.5 percentage points of winnability (90.48% down to 81.95%). If you are playing blind, like our bot, it costs you about three quarters of your wins (42.3% down to 10.6%).
That is the whole story in one sentence: Turn 3 does not remove many wins from the deal. It removes your ability to find them.
Floor, ceiling, and where you actually live
Think of three layers:
- The ceiling — what is possible if you know every hidden card (our solver, and Blake & Gent's 90.48% vs 81.95%).
- The floor — what a blind, no-undo player manages (our bot's 42.3% vs 10.6%).
- You — somewhere in between, and movable, because you remember cards, plan ahead, and can use Undo and Hint.
For real-world context, the site SolitaireX publishes its own player data. In Data Insights and Strategies to Improve Your Win Rate they report 913,106 games of classic Klondike played in a month, with 409,374 wins — an overall win rate of about 45% — and note that their strongest players sit far above that average. Real people, with memory and patience, beat any blind bot comfortably.
Why the gap is biggest for a player who cannot see the cards
At Turn 1 the stock is a menu. Every pass, you look at every card, and the order you take them in is yours to choose. At Turn 3, the stock is a menu you read three lines at a time, and only the bottom line is available. Two cards in three are invisible to you until something changes.
Three practical consequences:
- Your choices shrink. A Turn 1 player choosing between eight stock cards will often find the one useful move. A Turn 3 player sees the third card only and frequently sees nothing playable at all.
- Order becomes fixed. At Turn 3, the triples are locked together. If you cannot change the offset, you will keep seeing the same third cards forever — and the two cards you need stay buried.
- Mistakes cost more. Taking a stock card you did not need shifts the offset for the whole rest of the pass. A blind player cannot tell a good shift from a bad one, which is exactly why the bot collapses from 42.3% to 10.6%.
How to play Turn 3 better than a bot
The good news is that the bot's weaknesses are the exact things you can fix.
- Count the pass before you spend it. Run the stock through once and simply watch. Note which cards land as the third of each triple. That single habit is the biggest gap between you and a blind player.
- Use a stock card to change the offset on purpose. Playing one card shifts every later triple. If two cards you need are stuck in the middle of triples, that shift is how you free them.
- Turn face-down cards over first. Uncovering the tableau gives you new options for free. Filling a column with a King you do not need does not.
- Don't rush an Ace-to-foundation move if the card still has work to do. Low cards on the tableau catch other cards. Once they go up, they are gone.
- Prefer moves that open the longest buried column. Depth is where losses hide.
- Empty columns are tools, not trophies. Hold one open for a King you have actually seen, or to unload a stubborn pile.
Want to feel the difference yourself? Play a few hands of Klondike Solitaire at Turn 1, then switch to Klondike Turn 3 and notice how much more often the stock offers you nothing. Solitaire Draw 3 is the same three-card rule if that is the name you know it by.
How much the stock rule matters in other games
Draw rules only matter where there is a stock to draw from. A useful way to sort the games on this site:
Stock turned over and over
Klondike is the big one, and on this site it has unlimited redeals — we did not simulate a pass limit, so we give no Klondike pass-limit figure. Pyramid Solitaire is the only game where we measured a pass limit at all (one pass versus three), and we will publish those results on their own rather than mix them in here.
Stock used once
In Spider Solitaire the stock is dealt onto the tableau in rows and never comes back, so there is no draw-rule choice to compare — difficulty comes from the number of suits instead, which is why 2 suits and 4 suits are separate games. Forty Thieves, Easthaven and Golf Solitaire also give you one trip through the stock. Plan before you deal, because there is no second look.
No stock at all
Everything is face up from move one in FreeCell, Yukon, Scorpion and Simple Simon. There is no draw rule, so no Turn 1 versus Turn 3 comparison exists — and we have not measured one. These games reward pure planning: the information is all yours, the difficulty is deciding what to do with it.
You can browse the full list on our games page.
How to play here
Everything on solitaire-turn-3.com is completely free, with no ads of any kind, no pop-ups and no paywalls. There is nothing to download or install, and no sign-up, account or email address is needed. Open a game from the home page and the cards are already dealt.
- Large, clear cards and readable text, plus a zoom control to make the board bigger — comfortable for older players and anyone with tired eyes.
- Hint, unlimited Undo, Restart and Auto Play to finish a game you have already solved. Every feature is available to everyone.
- Drag and drop or simply tap to move cards, on desktop, tablet or phone in any modern browser.
- Klondike offers both Turn 1 and Turn 3 draw modes, so you can run the comparison in this article on your own deals.
- Wins, time and moves are tracked for the current game.
- Sound can be switched off, and fullscreen mode and table color settings are there if you want them.
No gambling, no in-app purchases, no timer pushing you along. The site exists for enjoyment — play one hand over coffee or ten on a rainy afternoon.
Frequently asked questions
Is Turn 3 solitaire really four times harder than Turn 1?
For a player who cannot see the hidden cards, roughly yes. On our 10,000 numbered deals, the blind bot won 42.3% (41.3–43.3%) at Turn 1 and 10.6% (10.0–11.2%) at Turn 3. For a player who knows every card, the gap is much smaller: Blake & Gent report 90.48% versus 81.95% winnable.
What win rate should I expect at Turn 3?
We cannot give you one number, and we will not invent one. Our bot figure is a floor and our solver figure is a ceiling; a thoughtful human with Undo lands in between. SolitaireX's published player data shows an overall win rate of about 45% across 913,106 classic games in a month, which is a helpful reminder that real players do far better than a blind bot.
Why do some solver results say "undecided"?
We gave the solver 150,000 positions per deal. When it neither found a win nor proved the deal dead within that budget, we record it as undecided — 21.3% (18.9–24.0%) of Turn 1 deals and 24.3% (21.7–27.1%) of Turn 3 deals. That is why the winnable share is a range rather than a single figure.
Does using Hint or Undo mean I am cheating?
No. There is no leaderboard to protect and nothing to buy. Undo is how you learn which stock offset was the wrong one, and Hint is there for when a board genuinely has no move. Use them freely.
How many passes through the stock do I get at Turn 3 here?
Klondike on this site has unlimited redeals. We did not simulate a pass limit for Klondike, so we do not publish a figure for one. The only pass limit we have measured anywhere is in Pyramid, comparing one pass with three.
Which game should I play if Turn 3 feels too punishing?
Switch to a game with no hidden stock. FreeCell puts every card in front of you, and Golf Solitaire gives quick, light hands. Or just flip Klondike back to Turn 1 — the same deals, four times more forgiving for a blind player.
Admin, for the solitaire-turn-3.com team.
Sources
- Charlie Blake & Ian Gent, The Winnability of Klondike Solitaire and Many Other Patience Games, Journal of Artificial Intelligence Research 85 (2026)
- SolitaireX.io. Data Insights and Strategies to Improve Your Win Rate
- JAIR article record: The Winnability of Klondike Solitaire and Many Other Patience Games