FreeCell: every card is face up, so what should you read first?
FreeCell is the one game on klondike.nz where nothing is hidden. All 52 cards are dealt face up into eight columns, which means the whole game is sitting in front of you before you make a move. That makes it perfect for our question on this site: what does the opening layout tell you?
We dealt 10,000 FreeCell games under this site's rules and inspected every one. Only 49.0% of deals had an Ace you could play on the very first move — so about half the time you start by digging. And yet our solver won 99.68% of the 10,000 deals, and did not find a single deal it could prove impossible. The remaining 0.32% were deals it could not finish within its search limit.
The lesson is simple. In FreeCell, a slow-looking start is almost never a lost cause. The layout tells you where to begin, not whether you can win.
The FreeCell deal, by the numbers
| What we looked at | Result | Deals | How we got it |
|---|---|---|---|
| Deals with an Ace playable on move one | 49.0% (95% range 48.0–49.9%) | 10,000 | Inspecting the deal |
| Deals the solver won | 99.68% (99.55–99.77%) | 10,000 | Best-first search, 200,000 positions, then 1,000,000 for any unfinished deal |
| Deals proven impossible | 0 | 10,000 | Same solver |
| Length of the first solution found | about 62 moves | 9,913 solved deals | Same solver, first pass |
| Published winnable share | 99.998881% | about 10 million | Blake & Gent, 2026 |
FreeCell as it is played here
Eight open columns
Eight columns: four of seven cards and four of six, all face up. Above them sit four free cells (each holds one card) and four foundations.
Finishing the game
Move every card to the foundations, by suit, from Ace to King.
Moves
- The bottom card of any column can go onto a card one rank higher of the opposite colour.
- Any single card can be parked in an empty free cell.
- Any card can go into an empty column.
- You can move a run of several cards at once; the game allows as many as your empty free cells and empty columns plus one. It is only a shortcut — the same move could be made card by card.
What to read in the first thirty seconds
Where are the Aces, and what is on top of them?
In about half of all deals (49.0%) an Ace is already free. In the other half, find the Ace with the fewest cards above it and count how many free cells you would need to reach it. That count is your opening plan.
Which column is shortest to clear?
An empty column is worth more than a free cell because any card can go there and it enlarges the runs you can move. Look at the four six-card columns first; one of them is usually the quickest to empty.
Where are the low cards trapped under Kings?
A 2 or 3 sitting under a King is a problem you will have to solve with space. Spot these early, before your cells fill up.
Plan for about sixty moves
The first solutions our solver found averaged about 62 moves. That is not necessarily the shortest route, but it is a good reminder that a winning FreeCell game is a long, steady sequence, not a flash of brilliance.
How much each free cell is worth
To see how important the cells are, we removed them one at a time and ran 3,000 deals for each setting:
| Free cells | Solved | Proven impossible | Undecided | Published winnable |
|---|---|---|---|---|
| 4 (as on this site) | 99.68% | 0% | 0.32% | 99.998881% |
| 3 | 96.0% | 1.0% | 3.0% | 99.3583% |
| 2 | 74.0% | 20.2% | 5.8% | 79.544% |
| 1 | 17.4% | 79.6% | 3.1% | 19.348% |
| 0 | 0.2% | 99.7% | 0.1% | 0.2137% |
Every published figure falls between our "solved" and "solved plus undecided" numbers. The practical takeaway: losing two cells turns a near-certain win into roughly three-in-four. Each cell you fill without a good reason is a real cost.
A first-minute checklist for FreeCell
Because every card is visible, a FreeCell game can be planned before you move anything. This is the routine we would suggest, built on what the deals in our study looked like.
- Spot any free Ace. About half the time (49.0% of our deals) one is already exposed. Send it home, and its 2 if that is exposed too.
- Rank the buried Aces. For each remaining Ace, count the cards on top of it. The one with the fewest is your first target.
- Pick the column you will empty. Usually one of the four six-card columns. Check that its cards have somewhere to go that doesn't use your last free cell.
- Note the blockers. Kings and Queens sitting on low cards are the cards that will eat your free cells. Plan where each one will go.
- Keep two cells in reserve for as long as you can. Our reduced-cell tests showed how quickly options shrink: with only two cells, the solver's success rate fell to 74.0%.
- Move, then re-read. After every few moves, run through the list again. The picture changes fast once a column opens.
None of this guarantees a win, but it matches what the numbers say: in FreeCell the deal is almost never the problem. Our solver didn't prove a single one of its 10,000 deals impossible, so if a game goes wrong, the answer is usually a few moves back — which is exactly what unlimited Undo is for.
Playing FreeCell on klondike.nz
It is free and works in any modern browser on computer, tablet or phone — nothing to download, no sign-up, no email. Cards are big and clear with a zoom control, and you have Hint, unlimited Undo, Restart and Auto Play. Drag cards or tap them. The game tracks wins, time and moves for the current game, and you can switch sound off, go fullscreen or change the table colour. Good company for a long drive's rest stop or a wet weekend at the bach.
Where these numbers come from
The klondike.nz team ran this study in September 2026 to measure what a FreeCell deal on this site gives you to work with.
- Rules from our code. Eight columns (7,7,7,7,6,6,6,6), four cells, alternating-colour building, any card to an empty column.
- A program copy of the game. The solver's proofs of impossibility were cross-checked with a completely separate brute-force search on a sample of deals (seeds 1–5 with one cell, and seeds 739 and 1053 with three cells); the two agreed on every deal tested.
- Numbered deals. Seeds 1 to 10,000 for four cells, 1 to 3,000 for each reduced setting. These are our own deal numbers, not the old Windows ones.
- Looking at each deal. We checked whether an Ace was playable immediately.
- The solver. A best-first search with a 200,000-position limit per deal; the 87 four-cell deals left unfinished were run again with 1,000,000 positions.
- Comparison. Every published figure sits inside our ranges.
- Error bars. Our 95% range for four cells, 99.55–99.77%, is where the true solver success rate very likely lies.
- Limits. This site lets a card come back off a foundation; the solver never does that, which can only make our figure slightly low.
Under the bonnet
Written in Python 3.11 with only the standard library, in a module called freecell.py (which also runs Baker's Solitaire), plus a retry.py script for the second pass. A shared runner uses multiprocessing to spread deals over 19 worker processes; the four-cell run took about 18 minutes. The solver always expands the most promising position next — the one with the fewest cards still to go home — keeps a table of positions already seen so none is searched twice, and makes only foundation moves that are provably safe. Ranges are Wilson score 95% intervals. We used an AI assistant, Claude by Anthropic, to help write and review the code and to draft this page. The figures were not produced by AI: they come from running the code, and we compared them with the published research before publishing.
Questions players ask
Is there such a thing as a bad FreeCell deal?
Very rarely. Our solver found no deal it could prove impossible in 10,000; the published estimate is that roughly 1 deal in 90,000 cannot be won.
What if no Ace is available at the start?
That is the case in about half of deals (49.0% have one available). Find the most reachable Ace and plan your cells around it.
Should I keep my free cells empty?
As far as you can. With two cells instead of four, our solver's success rate dropped to 74.0%.
How is FreeCell different from Baker's Solitaire?
Baker's builds in suit and, here, only lets Kings into empty columns, which makes it far harder. See our Baker's Solitaire page.
Who writes these pages?
The klondike.nz team. Read more on our about page, or head back to classic Klondike.
By Admin, klondike.nz. Study date: 18 September 2026.