Guide · Updated August 2026
Inaccuracy vs mistake vs blunder: what the labels mean
By Audit ChessIn chess, an inaccuracy is a small error that slightly worsens your position, a mistake is a clear error that gives away a meaningful advantage, and a blunder is a serious error that usually loses material or the game. Engines classify each by how much your win probability drops.
The three labels, by how much they cost
Chess engines grade your moves by how far they move the evaluation against you. An inaccuracy is a small slip - the position is slightly worse but nothing decisive has happened. A mistake is a clear error that hands over a meaningful advantage. A blunder is a serious error that typically loses material or turns a good position into a lost one.
The exact thresholds vary by tool, but the ordering is universal: inaccuracy is smaller than a mistake, which is smaller than a blunder. Some tools add a "miss" label for failing to punish an opponent's error, and top labels like "brilliant" or "great" for finding the only good move.
What the engine evaluation actually means
Engine evaluations are usually shown in pawns (centipawns): +1.0 means White is about a pawn better, -2.5 means Black is winning by roughly two and a half pawns of advantage, and a mate score like #3 means forced mate in three. The number is the engine's best guess at who stands better and by how much, assuming best play from both sides.
A move is judged by the change in evaluation, not its absolute value. Going from +5 to +3 is still winning, so it is at most an inaccuracy. Going from +1 to -2 flips the game and is a blunder even though no piece was captured.
Why win probability is fairer than raw pawns
A drop from +9 to +6 barely matters - you were winning easily and still are. A drop from +0.5 to -2.5 decides the game. To reflect this, modern tools convert the evaluation into a win probability and grade the swing in those terms, so errors are weighted by how much they actually change the result rather than by raw centipawns.
Accuracy scores work the same way: they summarise how closely your moves matched the engine's best line across the whole game, expressed as a single percentage per side. Club players commonly land in the 70s and 80s.
Practical examples
Inaccuracy: you retreat a bishop to a passive square instead of the active one. Nothing is lost, but you have conceded a little. Mistake: you allow a strong pawn break that gives your opponent a lasting initiative. Blunder: you leave a knight hanging, or walk into a fork that wins the exchange.
The useful question after each is the same: what was the move I should have played, and what pattern would have warned me? That is how a label becomes a lesson.
Finding the mistakes you repeat
One inaccuracy is noise. The same mistake in the same structure, over and over, is the thing to fix. Reviewing a single game shows you that game; reviewing your whole history shows you the pattern - which openings your accuracy drops in, which tactics you keep missing, which endings you misplay. Paste a game to see its labels move by move, or analyse your full history to find what repeats.
Keep reading
Frequently asked questions
What is the difference between a mistake and a blunder in chess?+
A mistake gives away a meaningful advantage; a blunder is more severe and usually loses material or the game outright. Both are bigger than an inaccuracy, which only slightly worsens your position. Engines separate them by how far the evaluation swings against you.
What does a +2 evaluation mean in chess?+
It means the engine thinks one side is better by about the value of two pawns, assuming best play. It is a measure of advantage, not a guarantee - the game still has to be converted over the board.
What is a good accuracy score in chess?+
Accuracy measures how closely your moves matched the engine's recommendations. Most club players score in the 70s and 80s per game; strong players are higher. A single low-accuracy game usually points to one or two decisive blunders.