How to Play Minesweeper

Minesweeper has a reputation for being unexplainable. People click around, hit a mine, restart, hit another mine, and conclude it's a luck game. It isn't. It's a deduction game with an interface that never bothered to teach itself.

The whole thing rests on one sentence: a number tells you how many mines are hiding in the eight cells touching it. Everything else follows from that.

This page walks through a single game from the first click to the last, narrating the reasoning at each step. Play along on the board above — that's a real game, not a video.

What you're looking at

Three things sit above the grid.

The mine counter, on the left. It starts at the board's total mine count — 10 on beginner — and drops by one each time you place a flag. It is counting *your flags*, not discoveries. Place ten flags on ten empty cells and it still reads 000. It can go negative if you over-flag. It's a bookkeeping aid, nothing more.

The reset button, in the middle. Click it any time for a fresh board. It also reports status: ↻ while playing, ✓ on a win, ✕ on a loss.

The timer, on the right. Starts on your first click, not on page load. Ignore it for now.

Each cell in the grid is in one of three states:

• Covered — untouched. Could be anything.
• Revealed — opened. Shows a number, or is blank if no mines touch it.
• Flagged — you've marked it as a suspected mine. It's still covered; the flag is your note, and the game never confirms it.

Left-click opens a cell. Right-click places or removes a flag. On a phone, tap to open and press-and-hold to flag.

Your first click, and why the corner

Click a corner.

The reason is about what happens next, and it's worth understanding because it shapes the rest of the game.

A cell's number counts mines among its neighbours, but not every cell has the same number of neighbours. A corner cell touches 3 cells. An edge cell touches 5. A cell in the middle touches 8.

When you open a cell that has zero adjacent mines, the game doesn't just open that one — it automatically opens everything connected to it, spreading outward until it hits cells that have numbers. That cascade is where you get your opening region. One click, dozens of cells revealed.

Centre clicks tend to cascade wider, since each cell has more neighbours to spread through. That sounds better, and for raw cells-opened it is. But it leaves you with a ragged frontier floating in the middle of the board, covered cells on every side.

A corner gives you a smaller region braced against two walls. Walls matter because they *limit where mines can be*. A 1 sitting against the edge has only 5 neighbours to worry about instead of 8, and if it's in the corner, only 3. Fewer possibilities means easier deductions. You want your early reasoning happening where the board is most constrained.

On this site there's also a guarantee: your first click is never a mine. The board doesn't even generate mines until after you click, and it places them so your first cell has no adjacent mines — meaning your first click always cascades. That's how the engine works. You cannot lose on move one here.

Reading the first numbers

Say you clicked the top-left corner and a region opened up. Somewhere along its border you find this:

   .   .   .
           .   1   ?
           .   ?   ?

The blanks are revealed cells with zero adjacent mines. The 1 is revealed and touching exactly one mine. The ? cells are still covered.

That 1 means: of the eight cells surrounding this one, exactly one contains a mine. Five of its neighbours are already revealed and blank, so they're not mines. Three are still covered. One of those three is a mine — you just don't know which yet.

Not enough to act on. Look elsewhere.

Now suppose you find this along the top edge:

   ▓   ▓   ▓   ▓
           .   1   ?   .
           .   .   .   .

Here ▓ is the board edge. That 1 is on the top row, so it has 5 neighbours instead of 8. Four are revealed and blank. Exactly one is covered.

The 1 needs one mine among its neighbours. Only one neighbour can possibly hold it. That covered cell is a mine. Certain — no guessing.

This is the entire game, and it never gets more complicated than this in principle. Compare a number against its covered neighbours. When the arithmetic leaves exactly one possibility, you've found something.

The mirror case is just as valuable. If a 1 has one covered neighbour and you've already flagged a mine elsewhere in its eight, then the 1 is satisfied, and every remaining covered neighbour is safe to open. Numbers tell you where mines are *and* where they aren't. Beginners chase the first and forget the second, which is why they run out of moves and start guessing.

Placing your first flag — and the trap

Right-click the cell you just proved is a mine. The counter drops from 010 to 009.

Now the thing nobody warns you about: flags do nothing.

A flag doesn't defuse anything. It doesn't score points. The game never tells you whether it's right. You could flag every cell on the board and be no closer to winning. And you win by opening every safe cell — flagging mines is not part of the win condition at all. You can finish a game having placed zero flags.

So why place any? For two reasons only:

1. Memory. A flag records a conclusion so you don't re-derive it every time you look at that corner.
2. Chording. Once a number has the right count of flags around it, you can open all its remaining neighbours in one action. That needs flags to work.

The mistake almost every new player makes is over-flagging — carefully flagging every mine they find, treating flags as progress. It feels productive. It's mostly wasted motion, and it's actively harmful when you flag a cell you only *suspect*. A wrong flag doesn't announce itself. It sits there and poisons every deduction that touches it, and you'll trust it, because you put it there.

The habit to build: flag only what you've proven, and mainly when the flag lets you chord. Otherwise just open the safe cells.

Playing out the game

Back to the board. You've flagged one mine. Look at what that unlocked:

   ▓   ▓   ▓   ▓
           .   1   F   1
           .   1   2   ?
           .   .   1   ?

The 1 on the top row, to the right of your flag, touches that flag. Its one mine is accounted for. So every other covered neighbour of that 1 is safe. Open them.

The 2 below touches your flag too. It needs two mines and has one. Its remaining covered neighbours hold exactly one more — not resolved yet, but you've narrowed it.

That's the rhythm of a game:

1. Scan the frontier — the border between revealed and covered cells. Nothing interesting happens in the middle of an open region.
2. Find a number whose covered-neighbour count equals its remaining mine count. Every one of those is a mine.
3. Find a number whose mines are all flagged. Every other covered neighbour is safe. Open them.
4. Opening cells reveals new numbers, which unlocks new deductions. Go back to 1.

Repeat until the board is clear. Each opened cell is information, and information compounds — one safe opening often unlocks three more deductions elsewhere.

When you get stuck

You'll hit positions where you scan the whole frontier and nothing resolves. Two honest possibilities:

You missed something. Most common. Re-scan, and this time check both directions — not just "which cells must be mines" but "which numbers are already satisfied, so their neighbours are safe." Beginners overwhelmingly miss the second kind.

The board genuinely can't be resolved. This is real. Minesweeper boards are generated randomly and aren't checked for solvability, so some positions require a guess. If you've truly exhausted the frontier, pick the cell with the best odds and click. Losing that click isn't a mistake — it's the game. There's a method for picking the least-bad guess, and the mine counter often turns an apparent coin flip into a certainty.

Before guessing, always check the counter. If it reads 001 and you have two covered cells that both look risky, one of them is safe.

Winning

When the last safe cell opens, the board is done — the reset button turns to ✓ and the timer stops. Any mines you never flagged are just left covered. You don't have to touch them.

If you hit a mine, the board reveals every mine's location, and any flag you placed on a non-mine gets marked as wrong. That endgame view is the most useful thing on a loss. Look at the wrong flags. That's where your reasoning broke, and it's usually one bad assumption that everything downstream inherited.

Where to go next

Play a few beginner boards. Aim to finish one without guessing — speed is irrelevant right now.

Once you're winning beginner consistently:

• Minesweeper rules — the exact rules as a reference, including board sizes and the ones people get wrong.
• Minesweeper strategy — the recurring shapes. The 1-2-1 pattern alone resolves positions that look impossible.
• How Minesweeper works — what the engine is doing: how the board is generated, why the first click is always safe, and why some boards are unwinnable.

The board at the top is set to beginner. Click a corner and start.