How to Play
The board is full of hidden mines. Tap a tile to open it. If it is a mine, the game ends and every mine is revealed. If it is safe, it shows a number from 0 to 8, and that number tells you exactly how many mines are hiding in the up-to-8 tiles touching it, diagonals included. A tile that shows nothing is a 0, meaning none of its neighbors are mines, so the game automatically opens all of them for you โ and keeps going outward through every 0 it finds, often clearing a whole region in a single tap.
Your very first tap of a new game is always safe and always opens a 0. The mines are not placed until you have clicked, and they are then placed anywhere except the tile you clicked and its eight neighbors, which guarantees that first tile reads 0 and starts a cascade. No run ever ends on the opening move, and no run ever begins by staring at a single isolated number.
Flag any tile you believe hides a mine: right-click on desktop, or press and hold on mobile. There is also a Flag Mode toggle in the bar above the board that turns every tap into a flag toggle, which is far more comfortable than holding each tile on a touch screen. Once a numbered tile has exactly as many flags around it as its number says, tapping that number again instantly opens all of its remaining unflagged neighbors โ this move is called a chord. You win by opening every tile that is not a mine. Flags are a bookkeeping aid and are not required to win; the mine counter simply shows total mines minus flags placed, and it will go negative if you over-flag.
Three difficulties are available at any time from the bar above the board: Easy (9x9, 10 mines), Medium (16x16, 40 mines), and Hard (20x20, 80 mines). The clock starts on your first tap, not when the board appears, and your best clear time is stored separately for each difficulty.
The Two Deductions That Solve Most of the Board
Minesweeper looks like a game about luck and is mostly a game about bookkeeping. The overwhelming majority of moves on any board follow from two rules, applied to one number at a time.
Start by rewriting every number in your head. A tile showing 3 with one flag already next to it is not a 3 any more โ it is a 2 among its remaining unopened neighbors. Subtracting the flags you have already placed is the single most useful mental habit in the game, and it turns most numbers into much smaller ones.
With that reduced number in hand, the two rules are:
- If the reduced number is 0, every remaining unopened neighbor is safe. All of that tile's mines are already accounted for, so everything else touching it can be opened without risk. This is exactly the condition a chord checks for you.
- If the reduced number equals the count of unopened neighbors, every one of them is a mine. A 3 with exactly three unopened neighbors left has nowhere else to put its mines. Flag all of them.
That is it. Cycle those two checks over the numbers along the boundary between opened and unopened territory, and boards unravel. Beginners lose games they could have solved because they scan the board looking for something interesting instead of walking the frontier mechanically. The interesting deductions are rare; the mechanical ones win games.
Comparing Two Numbers: The Subset Rule
When single-tile logic runs out, the next tool is comparing two numbers that share neighbors โ and it solves nearly everything the basic rules cannot.
Take two numbered tiles, A and B, and look at the unopened tiles each of them touches. If A's set of unopened neighbors is entirely contained inside B's, you can subtract one from the other. Whatever mines A accounts for are already inside B's territory, so B's remaining count minus A's remaining count must be distributed among only the tiles B touches that A does not.
Two useful conclusions fall out. If that difference is 0, every tile in B's set that A does not touch is safe โ open all of them. If the difference equals the number of those extra tiles, every one of them is a mine โ flag all of them. This is why two adjacent 1s so often let you open a tile several squares away from either of them: the first 1 already owns the shared region, so the second 1's outlying tile has nothing left to hide.
Learning to see this comparison quickly is the main difference between a player who solves 60% of boards and one who solves most of them without ever guessing after the opening.
Patterns Worth Memorizing
The subset rule produces the same handful of shapes over and over, and recognizing them on sight is much faster than deriving them each time. These are worth committing to memory.
1-2-1. Three numbers reading 1, 2, 1 sit in a row along the edge of your opened region, with three unopened tiles in a line directly against them and nothing else unopened nearby. The mines are under the two 1s, and the tile under the 2 is safe. The arithmetic is quick: the two 1s together account for two mines, the 2 in the middle also accounts for two, and the only arrangement satisfying all three puts a mine at each end and leaves the middle clean.
1-2-2-1. Four numbers reading 1, 2, 2, 1 in a row with four unopened tiles against them. Here the answer inverts: the mines are under the two 2s, and the tiles under the two 1s are safe. Players who half-remember 1-2-1 routinely misapply it here and open a mine, so it is worth learning the pair together rather than either alone โ same shape family, opposite answer.
1-1 against a wall. Two 1s side by side, where the first one's unopened neighbors are all shared with the second. The first 1's mine is somewhere in the shared pair, so the second 1's extra tile โ the one sticking out past the first โ is safe. This is the subset rule in its smallest form and it appears constantly along the board edges, which is one of several reasons edges are easier territory than the open middle.
Edges and corners help in general because they have fewer neighbors. An interior tile touches 8 unknowns, an edge tile 5, and a corner tile 3. Fewer unknowns means fewer possible arrangements, which means a number there constrains its surroundings far more tightly. Working the boundary of the board is not superstition; it is genuinely a smaller search space.
Chording, and Why a Wrong Flag Is Dangerous
Chording is the speed technique. Once a number's flag count matches the number itself, tapping that number opens every remaining unflagged neighbor at once. Instead of six separate taps around a satisfied 2 you make one, and each chord frequently triggers cascades that clear large regions for free. At any level of play above casual, most of your clicks should be chords.
This is also the reason flag discipline matters far more than it first appears. A chord does not verify your flags โ it simply trusts them. It counts the flags around the number, and if the count matches, it opens everything else. If one of those flags is on a safe tile, the mine it should have covered is sitting in the set the chord is about to open, and you lose the game on the spot. A wrong flag is not a harmless note to yourself; it is a loaded trap you set for your own next click.
So flag only what you have actually deduced. If you merely suspect a tile, leave it alone and open somewhere else โ an unflagged tile costs you nothing, since flags are not required to win. Some fast players go further and flag almost nothing, opening only tiles they have proven safe. That style gives up chording entirely in exchange for never being able to poison one, and it is a legitimate trade, especially on Easy where the deductions are short.
When There Is No Safe Move
Sooner or later you will exhaust every deduction and still have unopened tiles. Guessing is part of the game, but guessing well is a skill, and it has a clear structure.
Split the unopened tiles into two groups. The frontier is every unopened tile touching a revealed number; those tiles have real constraints on them, and the arrangements consistent with the numbers give some of them a much better or worse chance than others. The interior is everything else โ tiles far from any number, which are constrained only by the global count of mines still left. For interior tiles, the chance any one hides a mine is simply the remaining mines divided by the remaining unopened tiles.
The correct procedure is to work out the best frontier tile you can identify and compare it against that global rate. Frontier tiles are frequently better โ a 1 touching four unknowns puts each at roughly one in four, which beats a 20% board rate only slightly, while a 1 touching six unknowns is considerably better. But frontier tiles can also be much worse, and when every one of them looks bad, an interior tile at the board average is the right click.
Two practical adjustments. First, when two candidates are close in probability, prefer the one that will tell you more: a tile deep in unopened territory can cascade and hand you a whole new region, while a tile jammed against your existing frontier tends to reveal one number and leave you guessing again. Second, guess early rather than late. A forced guess when 60 mines remain among 200 tiles is a much better bet than the same guess deferred until 5 mines hide among 8 tiles, and boards where you postpone every uncertain click tend to end with the worst possible odds. Corners are worth a special mention here for the same reason edges are easy โ with only three neighbors, a corner guess that survives produces an unusually well-constrained number.
Finally, accept that some boards are genuinely unsolvable without a coin flip. It is a known property of the rules, not a flaw in your reasoning. Losing a board to a true 50/50 after playing it correctly is a different event from losing one to a missed deduction, and only the second kind is worth reviewing.
What Easy, Medium, and Hard Actually Change
The three difficulties are not simply bigger boards. Their mine densities differ, and density is what changes how the game feels.
Easy packs 10 mines into 81 tiles, a little over 12%. Medium runs 40 mines across 256 tiles, around 15.6%. Hard puts 80 mines into 400 tiles, exactly 20%. Those gaps sound small and are not. Density controls how common 0-tiles are, and 0-tiles are what produce cascades. At Easy density, your opening tap frequently strips away a third of the board and hands you a long, generous frontier full of low numbers. At 20%, cascades are noticeably tighter, openings are smaller, and you spend most of the game working a long boundary made of 2s and 3s, where deductions are longer and forced guesses arrive more often.
Density also changes the value of flagging. On Easy, with mines sparse, most numbers are 1s and 2s that resolve in a move or two, and heavy flagging can slow you down. On Hard, the numbers are large enough that tracking them without flags becomes genuinely difficult, and chording is close to mandatory if you want a respectable clear time.
Tips
- Subtract the flags first. A 3 with two flags beside it is a 1, and thinking of it as a 1 makes the next move obvious.
- Walk the frontier methodically instead of hunting for interesting spots. Most winning moves are boring ones.
- Chord constantly, but flag only what you have proven. A chord trusts your flags without checking them.
- Learn 1-2-1 and 1-2-2-1 as a pair โ the mines sit under the 1s in the first and under the 2s in the second.
- Work edges and corners when you have a choice: 3 or 5 unknown neighbors constrain a number far more tightly than 8 do.
- When forced to guess, compare your best frontier tile against remaining mines divided by remaining tiles, and take whichever is better.
- Guess early while the odds are good rather than saving all your uncertainty for the endgame.
- On Hard's 20x20 board, turn on Flag Mode for touch play instead of long-pressing every tile.
About This Minesweeper
This is a free online version of the classic Minesweeper logic puzzle, built to play equally well with a mouse and with a thumb. It runs in the browser with no download and no account.
Instead of the traditional 30x16 Expert board, Hard uses a 20x20 grid with a comparable mine density of 20% โ wide rectangular boards get uncomfortably cramped on phone screens, while a square grid stays readable at any size. Every board is generated fresh at the moment of your first tap, and because the mines are placed around that tap rather than before it, your opening is always safe and always cascades.
Your fastest clear time is saved locally in your browser for each difficulty separately, so switching difficulties will not overwrite your Easy record with a Medium one. There is no move limit and no penalty for taking your time โ only your best clear time is tracked, so you can play a slow, careful first pass and push for speed once the patterns become automatic.