Mines game odds: how the multiplier is calculated
How Mines works on a 5×5 grid: the chance of each safe pick, the multiplier formula, tables for 1 to 10 mines, and why cashing out early or late costs the same.
Fact-checked 18 October 2026 · 5 sourcesMines hides between 1 and 24 mines in a 5×5 grid. Every safe tile you reveal raises the multiplier, and you can cash out after any of them; reveal a mine and the stake is lost. The multiplier isn't arbitrary: it's the inverse of your chance of getting that far, trimmed by the house edge.
The rules
You choose how many mines to hide and place your bet. At that moment the mines are drawn: the bet's random floats pick tiles one at a time from those still left until every mine is placed. They don't move after that, whatever you click.
Then you reveal tiles one by one. Each safe tile sets the multiplier to (1 − edge) / P, where P is the chance that all the tiles you've revealed so far would be safe. Cash out after at least one safe tile and you're paid stake × multiplier; hit a mine and the round ends with nothing. An auto bet picks its tiles up front and pays only if all of them are safe.
The grid looks like Minesweeper, but the two games differ where it matters. In Minesweeper, opened cells show how many mines touch them, so you can work out which cells are safe. Mines gives no clues: every hidden tile has the same chance of being a mine, so corners, patterns and lucky tiles make no difference.
The chance of k safe picks
With m mines there are 25 − m safe tiles. Your first pick is safe with chance (25 − m) / 25. If it was, the next is safe with chance (24 − m) / 24, because one safe tile is gone, and so on. Multiply the steps together and you get
P(k safe picks) = C(25 − m, k) / C(25, k)
where C(n, k) counts the ways to choose k tiles out of n. This is the hypergeometric distribution, the math of drawing without putting anything back, in the case where every draw is a success.
With 3 mines, the first pick is safe 88% of the time (22/25), two in a row 77% (22/25 × 21/24), and five in a row 49.57%. Each step is a little riskier than the one before, because the safe tiles run out while the mines stay: 88%, then 87.5%, then 86.96%.
The multiplier
Divide 1 − edge by that chance and you have the multiplier. At a 1% edge, for example, one safe pick with 3 mines pays 0.99 / 0.88 = 1.125×, and one safe pick with 24 mines pays 0.99 / 0.04 = 24.75×. Multipliers are rounded down to four decimal places.
Chance that every pick is safe:
| Safe picks | 1 mine | 3 mines | 5 mines | 10 mines | |--:|--:|--:|--:|--:| | 1 | 96.00% | 88.00% | 80.00% | 60.00% | | 2 | 92.00% | 77.00% | 63.33% | 35.00% | | 3 | 88.00% | 66.96% | 49.57% | 19.78% | | 5 | 80.00% | 49.57% | 29.18% | 5.65% | | 10 | 60.00% | 19.78% | 5.65% | 0.092% |
Multiplier at a 1% edge:
| Safe picks | 1 mine | 3 mines | 5 mines | 10 mines | |--:|--:|--:|--:|--:| | 1 | 1.0312× | 1.125× | 1.2375× | 1.65× | | 2 | 1.076× | 1.2857× | 1.5631× | 2.8285× | | 3 | 1.125× | 1.4785× | 1.9973× | 5.0043× | | 5 | 1.2375× | 1.9973× | 3.3925× | 17.5153× | | 10 | 1.65× | 5.0043× | 17.5153× | 1,077.6131× |
The tables have a tidy symmetry: 3 mines with 5 picks is the same bet as 5 mines with 3 picks, because C(25 − m, k) / C(25, k) stays the same when you swap m and k. The game page shows the edge in force; put 1 − edge in place of 0.99 to redo the second table.
Why cashing out early or late costs the same
Stop after k picks and you win (1 − edge) / P with chance P. The expected return is P × (1 − edge) / P = 1 − edge, wherever you stop. Cashing out after one tile or after ten changes how often you win and how much, not what the game keeps on average.
You can see it step by step too. After the first tile, each new safe tile multiplies your multiplier by one over the chance that tile was safe: with 3 mines and two tiles already open, the third pick is safe with chance 20/23 and raises the multiplier by 23/20. So every reveal after the first is a fair bet on what you've built up, and the edge is charged once per round, on the stake. The cost of Mines grows with how many rounds you play and how much you stake, not with how many tiles you open. House edge and RTP shows how that charge adds up over many rounds.
None of this makes an early cash-out safer in the long run, or a late one smarter. Raising the stake after a lost round to win it back doesn't change the average either, which Martingale and other betting systems works through with numbers.
Big multipliers are rare by design
Clearing the whole board means finding every safe tile, with chance 1 / C(25, m). With 3 mines that's 1 in 2,300, paying 2,277× at a 1% edge. With 10 mines it's 1 in 3,268,760, and the longest odds of all come at 12 or 13 mines: 1 in 5,200,300. Numbers like these are why the top multipliers look so large.
Whatever the multiplier says, a single bet can't pay more than the game's maximum win, which the page lists with its other limits. At high mine counts and long runs, that cap can sit well below stake × multiplier, so it's worth checking before you aim for a full board.
Checking a round
Because the mines are drawn from the bet's seed pair, every round can be replayed. The floats come from HMAC-SHA256(server seed, "client seed:nonce:k"), eight to a block, and each one picks a tile from those still left: floor(float × tiles left). After you rotate your seed pair the old server seed is revealed, and the provably fair calculator redraws the grid so you can compare it with the round you played. How to verify a provably fair bet goes through the same check by hand, and the game itself is at Mines.
