FreeCell Two Decks: bigger board, still a puzzle you can solve
FreeCell Two Decks takes everything that makes FreeCell satisfying (every card face up, free cells to park cards, no luck of the draw) and doubles it: 104 cards, ten columns, eight free cells and eight foundations. How winnable is it? Our solver took on 400 deals and solved 88.0% of them. It did not find a single deal that was impossible; the other 12.0% ran past its search budget. So the honest answer is that somewhere between 88% and 100% of deals can be won, and the evidence leans towards the top of that range.
It is a long game. The solver's first solutions averaged about 148 moves, so this is one for a wet weekend at the bach rather than a quick break.
At a glance: FreeCell Two Decks in numbers
| What we measured | Result | Deals | How |
|---|---|---|---|
| Deals solved | 88.0% (95% range 84.4–90.8%) | 400 | Solver with every card visible, 150,000 positions per deal |
| Deals proven impossible | 0 (95% range 0–0.95%) | 400 | Same solver |
| Deals left undecided | 12.0% | 400 | Ran past the search budget |
| Length of the first solution found | about 148 moves | 352 solved deals | Not necessarily the shortest solution |
| Deals with an Ace at the bottom of a column | 55% | 400 | Counted from the deal |
| For comparison: one-deck FreeCell solved | 99.68%, none impossible | 10,000 | Same solver design; see our FreeCell page |
The rules on this site
- Two packs, 104 cards, all dealt face up.
- Ten tableau columns: four of 11 cards and six of 10.
- Eight free cells, each holding one card.
- Eight foundations (a pair for each suit), each climbing from Ace to King.
- On the tableau you build down in alternating colours: a red 9 on a black 10, a black 4 on a red 5.
- Any card may go into an empty column.
- You can move a run of several cards at once when you have room to do it. The most you can move in one go is empty free cells + empty columns + 1. With enough space you could always shift the cards one by one instead, so this limit is a convenience, not a new rule.
What the numbers tell you
Losing is almost never the deal's fault
Out of 400 deals, not one was proven impossible. Our solver ran out of budget on 48 of them. In one-deck FreeCell, where we could afford to re-run undecided deals with a much bigger budget, that extra search never turned up an impossible deal. If a game of Two Decks beats you, assume there was a way through.
A bigger board means bigger plans
The first solutions our solver found averaged about 148 moves, compared with about 62 for one-deck FreeCell. You are not just playing twice as many cards; you are juggling more runs at once. Pace yourself.
Aces are often close at hand
In 55% of deals at least one Ace already sits at the bottom of a column, ready to go up on your first move. With eight Aces to find, though, some will always be buried deep in the eleven-card columns.
Strategy for two decks
Eight cells are generous; keep them that way
Eight free cells feel like plenty at the start. They fill up quickly on a ten-column board. Treat them as working space for the next few moves, not long-term storage.
Empty a short column early
The six ten-card columns are slightly easier to clear than the eleven-card ones. An empty column is worth more than a free cell: it can hold a whole run, and it adds to the number of cards you can move at once.
Use the duplicate cards
Every card has a twin. If you need a black 6 to go on a red 7, there are four of them in play (two spades and two clubs), and any one will do on the tableau. Look for the easiest one to reach.
Keep paired foundations level
Two foundations per suit give you flexibility, but only if both keep moving. Racing one clubs pile to 9 while the other sits on 2 can strand low clubs in the tableau.
Plan in chunks
Pick one area of the board (two or three columns) and sort it before moving on. On a 104-card layout, trying to track everything at once is how mistakes creep in.
Playing FreeCell Two Decks here
It is free to play in any modern browser on a computer, tablet or phone, with nothing to install and no account or email needed. The large cards and zoom control help a lot on a ten-column board. You get Hint, unlimited Undo, Restart and Auto Play, drag-and-drop or tap-to-move, and a count of wins, time and moves for the current game. Sound can be muted, and fullscreen and table colours are in the settings.
How we did the research, step by step
The free-solitaire.nz team ran this study in September 2026 to measure how winnable FreeCell Two Decks is under this site's own rules.
- Read the rules from the site's code. The game is the "two decks" branch of the site's FreeCell engine: 104 cards, ten columns (four of 11, six of 10), eight free cells, eight foundations, alternating-colour building, any card in an empty column, and a multi-card move limit of empty cells + empty columns + 1.
- Rebuilt the game as a program and checked that it only makes the moves those rules allow.
- Dealt numbered games. Deals 101 to 500. Deals 1 to 8 had been used to tune the solver's scoring, and testing on the same deals would flatter it, so the measured run starts well clear of them.
- Chose a solver. Every card is face up in this game, so a solver sees exactly what you see. It either finds a win, proves there is none, or runs out of budget ("undecided").
- Ran it with a budget of 150,000 positions per deal. The 400 deals took about 11 minutes.
- Looked for published figures. The Blake and Gent study (2026, Journal of Artificial Intelligence Research) does not include two-deck FreeCell, so there is nothing to compare with directly. Our one-deck FreeCell results do agree with that study.
- Worked out the margin of error. The 95% range for "solved" is 84.4–90.8%: if we ran 400 fresh deals, the true share would land in that range 95 times out of 100.
- Stated the limits. 400 deals is a smaller sample than our other games, chosen to keep the run to a reasonable time, and 12% of deals were undecided. That is why we quote a range rather than one number.
The technical side
The program is Python 3.11 with nothing but the standard library. FreeCell Two Decks shares a module, freecell2.py, with Eight Off. The solver is a best-first search: it always explores the most promising position next, keeps a table of positions already seen so none is searched twice, and makes provably safe foundation moves automatically. A shared runner.py spreads the deals across parallel worker processes with multiprocessing, and every deal is shuffled from its own seed so it can be replayed exactly. Each 95% range is a Wilson score interval. We had help from an AI assistant (Claude, made by Anthropic) in writing and reviewing the code and in drafting this page, but no figure here was produced by the AI: they all come from running the solver, and the team checked them before publishing.
By Admin, free-solitaire.nz. Study date: 19 September 2026.
Frequently asked questions
Is there such a thing as an unwinnable Two Decks deal?
We did not find one that cannot. Our solver won 88.0% of 400 deals outright and proved none impossible; the rest were undecided.
Is it harder than normal FreeCell?
It is longer and needs more planning, but it is not obviously less winnable. The main difference is scale: about 148 moves per solution against about 62.
Why eight free cells?
Two packs means twice as many cards, so the game doubles the parking spaces to keep it fair.
Can I put any card in an empty column?
Yes, any card or a run of cards, subject to the move limit.
What should I try next?
Try Eight Off for a different kind of cell game, or go back to classic FreeCell.