Inaccuracy, mistake and blunder: what is the difference?

Updated 7 November 2026 · By ChessGrade

In one documented classification, inaccuracy, mistake and blunder indicate increasingly serious losses according to that provider’s grading model. One provider describes an inaccuracy as a weak move, a mistake as a bad move that immediately worsens the position, and a blunder as a very bad move that loses material or the game. Those descriptions are specific to that provider: the calculation and boundaries can differ between systems. Our own suggestion is to treat the word as an invitation to look at the particular position, not as a universal verdict on your ability.

Does an inaccuracy mean the move is always bad?

In one documented grading model, an inaccuracy describes a loss of expected points within that model’s inaccuracy range. In one published classification model, an inaccuracy has a smaller loss of expected points than a mistake or blunder. That provider's expected-points model uses the player's rating and the engine evaluation to estimate likely outcomes before and after the move.

Our own suggestion: when this label surprises you, ask which alternative the analysis preferred and what consequence it considered. Instead of treating the word as proof that the move was obviously wrong, compare its intended idea with the alternative. The label can be a starting point for a question, rather than the answer to every question about the position.

Our own suggestion is to describe that purpose in ordinary language: for example, were you trying to protect a piece or make a threat? Then frame the issue as whether a different choice would have served that purpose more effectively. These are questions for discussion, not a claim that any particular move or plan is sound.

What makes a mistake different from an inaccuracy?

In the published model described above, mistake represents a larger loss of expected points than inaccuracy. The provider's explanatory page describes a mistake as a bad move that immediately worsens the position. Neither description establishes a single definition that every chess tool must follow.

Our own suggestion: consider the label alongside the specific consequence shown for the move. The useful question is not just “Was this called a mistake?” but “What changed in this position?” If the alternative is unclear, the label alone does not explain your thought process or why you selected your move. It can be useful to write your own explanation before treating the label as a complete account.

The terms describe the move under a system's criteria, rather than ranking the person who played it. Our own suggestion is to keep those two things separate: a move can receive a negative label without telling you everything about the rest of the game. A single word also leaves room for a follow-up question about the particular alternative and its consequences.

Is a blunder always a lost game?

The provider's short description says a blunder can lose material or the game. In its published expected-points model, blunder occupies the largest loss band of the three labels. That does not establish a promise that every move called a blunder by every system immediately loses the game.

Our own suggestion: distinguish the stated consequence from the label itself. If a move is called a blunder, ask what the analysis is assuming the other player can do next. That question is more specific than simply asking whether the move deserves an alarming word. If the answer remains unclear, bring the position and the suggested continuation to a coach or another player for discussion.

Why can the same move get a different label?

A chess engine's maintainers explain that their engine supplies evaluations and preferred moves, while websites and graphical interfaces often add annotations separately. The engine output and the label are therefore separate things in that documented setup.

Different systems need not assign the same category to a move, because the label depends on the classification method as well as the analysis feeding it. Our own suggestion: before comparing two labels, check whether you are looking at the same position and the same move. A difference in the displayed word is a reason to investigate the method and the suggested continuation; it is not, by itself, a demonstration that either result is infallible.

The public documentation for one model bases its categories on a change in expected points. This is a description of that model, not a universal numerical threshold. Even where a provider publishes cutoffs, applying those cutoffs to a different system without evidence would misrepresent what that other system is doing.

What should I take away from a surprising label?

Our own suggestion is to turn the label into a small set of questions: what was the idea behind my move, what alternative was preferred, and what consequence distinguishes the two? Keep the questions tied to that one position instead of trying to infer a complete assessment of your chess from a single label. If the answer is still not clear, it is reasonable to leave the issue open rather than invent an explanation.

Our own view: these labels can be helpful shorthand, but they are not interchangeable with an explanation in plain language. A useful explanation connects the move, the alternative, and the relevant consequence. The provider-specific definitions above give you a vocabulary to start that conversation; they do not guarantee that every tool will use the same vocabulary in the same way.