Minesweeper Strategy

Open the online Minesweeper solver Bring a board, then compare its clues with these examples.

Minesweeper Strategy in One Page

Your first objective is to avoid guesses that a careful count could eliminate. Work along the boundary between open and covered cells. For each number, count its adjacent established mines and remaining covered neighbors, including diagonals. If its mine requirement is already satisfied, the remaining neighbors are safe. If it still needs as many mines as it has covered neighbors, those neighbors are all mines.

When one clue does not settle a move, compare it with the next. Shared neighborhoods often give a conclusion that neither number provides alone. Continue until new reveals stop producing deductions, then check the total mines and look for another frontier. Only after that should you compare uncertain choices. The goal is not to make uncertainty disappear by confidence; it is to know when your next click has a proof and when it carries risk.

Flags help record conclusions, but an unproved flag is an assumption. Keep that distinction clear, especially before chording: opening several neighbors because the flag count matches a number is safe only when the flags occupy the correct squares. One wrong flag can make a seemingly routine action lose the game.

How to Start a Board

Check the version's opening rule before choosing a starting square. In a new game generated on this site, mine placement protects the first clicked cell and its surrounding neighbors. That gives you a blank opening. It does not apply as an extra promise to a screenshot you import from an already running game, and other implementations can use different rules.

A corner is a convenient place to begin because the nearby boundary is easy to follow. It is not universally the best opening, and fewer neighbors do not by themselves prove that a corner will clear more squares. Opening size depends on mine placement and the rules used to generate the board. Pick a consistent starting habit, then judge your progress by the deductions you make afterward.

After the opening, scan the numbered edge rather than clicking isolated covered squares at random. A safe reveal beside several clues may supply useful information about their shared neighbors. However, information value comes after safety when a proved move is available: a promising-looking location is not worth an unnecessary risk merely because it might open a large area.

Guaranteed Safe Moves vs. Probability

A cell is guaranteed safe if every mine layout compatible with the visible information leaves it empty. This does not require a unique solution for the whole board. Several different layouts can agree about one safe square. Likewise, a certain mine is present in every compatible layout. Once you establish either result, use it to update nearby counts.

Try the 1-2-1 position below. Its middle covered cell is safe and the two outer cells contain mines, provided there are no additional unresolved neighbors outside the frame. Click a square to check the conclusion. The full Minesweeper patterns guide explains this comparison and four other useful shapes, with the boundary assumptions shown for each.

A probability answers a different question: how much of the remaining uncertainty puts a mine in this square? Under equally likely complete layouts, count the layouts containing a mine there and divide by the total number of compatible layouts. One mine among five truly equivalent choices gives each a 20% mine risk. Without that equivalence, dividing one by five does not establish the individual risks.

The current homepage tool offers logical suggestions rather than a percentage on every cell. Peek colors identify analysis results, not a scale from low to high risk. A missing hint is not an instruction to click, and it does not establish 50/50 odds. When comparing your own reasoning with the tool, read the suggested action and the visible clues together.

Counting Mines in the Endgame

Use the original mine total as a global constraint. Subtract the mines you have actually established, then ask how many of the remaining mines must belong to each unresolved region. Avoid adding the requirements of neighboring numbers directly: those numbers can be counting the same mine. The quantity you need is the number of distinct mines in the region, not the sum of its printed clues.

For example, suppose the game began with ten mines and you have proved seven. Three remain. If the entire unresolved frontier must contain exactly two mines, the covered squares away from all clues contain exactly one more. If there are three such squares and no other information distinguishes them, symmetry gives each one-third mine risk. If the frontier instead requires all three remaining mines, every square outside it is safe.

The displayed counter commonly subtracts placed flags, including wrong ones. Recheck any uncertain flags before treating it as evidence. Also check whether a region's mine count is fixed: different valid arrangements can use different totals there. You cannot subtract a single chosen arrangement's count and assume it represents every possibility. The untouched area and the frontier are connected by the global total.

When You Have to Guess

The small board below has two compatible layouts: its one mine is in A2 or in B2. Both arrangements satisfy both revealed 1s, and there are no other clues or covered cells outside this example. With symmetric generation and equal weighting, each candidate has a 50% mine risk. Clicking either square in the exercise confirms that the starting clues do not prove it safe.

In a full game, check whether another region can give you information before calling a pair isolated. A nearby safe reveal or the total mine count can sometimes distinguish alternatives that look identical locally. Conversely, changing the order of unrelated clicks will not resolve a truly isolated pair. Recognizing a real guess means accepting the information limit rather than searching indefinitely for a nonexistent clue.

1. Take known safe moves first. New clues can change both your deductions and the risks of later choices.
2. Compare supported risks. Use complete constraints rather than the number of unresolved clues a square happens to touch.
3. Consider useful information when risks are equal. A safe result that exposes a new boundary may help the rest of the game more than one that tells you little.
4. Match the choice to your goal. Maximizing survival on the next click, maximizing the chance of winning, and minimizing time spent on losing games are related but different objectives.

Guessing earlier solely because an early loss feels cheaper is a time-management preference, not a proof of better winning odds. Similarly, the square touching the fewest clues is not automatically the safest: clues can concentrate mines elsewhere and make a heavily constrained square better. Write down what supports your comparison instead of replacing one visual guess with another.

Strategy by Board Size: 9x9, 16x16 and 30x16

On beginner boards, slow down enough to explain each flag. A short board makes it easier to revisit the entire boundary after every reveal. On intermediate boards, make a deliberate circuit of the open regions so you do not guess beside one cluster while a safe move waits in another. Expert boards reward the same habits across a larger, denser field; their numbers do not follow different rules.

Change only one training goal at a time. If you are learning shared-neighbor deductions, ignore the timer and review uncertain positions. Once the reasoning feels reliable, practice efficient input and chording. Speed from fewer unnecessary actions is more useful than speed from clicking before you have checked the diagonal neighbors.

Check Your Board with the Solver

Open the online Minesweeper solver to practice a new game, or import a screenshot of a position you want to study. Recognition starts automatically. Review the cell types, correct mistakes, confirm the original total mines, and press Import. An accurate input matters more than the strength of an analysis method.

Use Step to inspect one suggested reveal or flag without changing the board. Compare it with your own reasoning and follow the solver walkthrough for the other controls. If you are continuing a game on another website, make a proved move in that original game and capture its new clues before analyzing again. A reconstructed local board cannot tell you the original locations of mines that the screenshot never revealed.

Try your board in the solver Check a safe move or a mine to flag, one step at a time.