Minesweeper Probability & Logic — The Math Behind Every Click

Minesweeper probability helps you make better decisions by understanding the mathematical odds behind every play.

The Probability Landscape

Every covered square in Minesweeper has a probability of being a mine. At the start, before any clicks, every square has the same probability: total mines ÷ total squares.

Once you start revealing squares, probabilities diverge. Some squares become certain (mine or safe), while others shift based on revealed numbers.

Unconditional vs. Conditional Probability

Unconditional (Prior) Probability

Before any information: probability = mines ÷ squares.

For Expert: 99 ÷ 480 ≈ 0.206.

Conditional (Posterior) Probability

After revealing numbers, each covered square’s probability updates based on:

• Which numbers can see it.
• What constraints those numbers impose.
• What other squares those numbers also see.

A covered square next to a “1” that already has one adjacent flag has a mine probability of 0 (it’s safe). A covered square that’s the only unflagged neighbor of a “1” has a mine probability of 1 (it’s a mine).

Most squares fall somewhere between these extremes.

Constraint Logic

Each number is a constraint:

$$\text{Number} = \sum(\text{mines among its covered neighbors})$$

When multiple numbers share covered neighbors, their constraints combine. This is a system of linear inequalities over binary variables (each square is 0 for safe or 1 for mine).

Example: Two Overlapping Constraints

Squares A, B, C are covered. Number X sees A and B. Number Y sees B and C.

• X = 1 → A + B = 1
• Y = 1 → B + C = 1

Subtracting: A − C = 0, so A = C. If further information reveals A = 0, then C = 0 and B = 1.

Enumeration Method

When constraints don’t resolve cleanly, enumerate all valid configurations:

1. List every covered square adjacent to a number.
2. Try every possible mine/safe assignment.
3. Discard assignments that violate any constraint.
4. Count how many valid assignments make each square a mine.
5. Probability = (mine assignments for square) ÷ (total valid assignments).

Example

Three covered squares, two constraints. Suppose valid configurations are:

• P(A = mine) = 1/2 = 50%
• P(B = mine) = 1/2 = 50%
• P(C = mine) = 2/2 = 100%

C is definitely a mine. A and B are 50/50.

The Global Constraint

The mine counter gives you a global constraint: the total number of unflagged mines remaining. This connects otherwise independent regions of the board.

Example: Two isolated regions remain. Region 1 has 5 covered squares. Region 2 has 3 covered squares. 4 mines remain total.

If analysis shows Region 1 must contain 2 or 3 mines, then Region 2 must contain 2 or 1 mine respectively. This cross-region inference can resolve squares that local analysis alone cannot.

Interior vs. Border Probability

Covered squares fall into two categories:

• Border squares — adjacent to at least one revealed number. Their probability is determined by local constraints.
• Interior squares — not adjacent to any number. Their probability equals the “leftover” density: (remaining mines not accounted for by border analysis) ÷ (number of interior squares).

In many positions, interior squares are actually safer than border squares, because local constraints can concentrate mines along the border.

Optimal Guessing

When you must guess, the mathematically optimal strategy is:

1. Calculate the probability of each candidate square being a mine (or estimate it).
2. Click the square with the lowest mine probability.
3. Prefer squares that reveal the most information — even if two squares have equal probability, the one adjacent to more unsatisfied numbers is a better guess because a safe result unlocks more deductions.

Expected Value

If square A has a 20% mine chance and square B has a 35% mine chance, always click A. Over many games, consistently choosing lower-probability guesses compounds into a significantly higher win rate.

The 50/50 Problem

A true 50/50 is when two squares each have exactly 50% mine probability and no further information can distinguish them. Common configurations:

• Two unconstrained squares at a corner.
• Two squares that symmetrically satisfy all nearby constraints.

50/50s are unavoidable. They’re the primary reason even perfect play doesn’t guarantee a 100% Expert win rate.

Reducing 50/50 Impact

• Solve the rest of the board first. Sometimes what appears to be a 50/50 resolves when other regions are cleared and the global mine count constrains the situation.
• Don’t create 50/50s. Choosing which area to work on first can sometimes avoid ending up in a 50/50 configuration.

Win Rate Analysis

Theoretical win rates for perfect logical play (optimal guessing, no misclicks):

The gap between your actual win rate and these numbers represents room for improvement in either logic or guessing decisions.

Summary

Minesweeper is a game of conditional probability. Every number updates the probabilities of its neighbors. Combining constraints narrows the possibilities. When logic runs out, the player with the best probability intuition wins the most games over time.

Ready to play? Try Minesweeper for free on Rare Pike — no download needed.