What Is a Good Accuracy in Chess? Typical Accuracy by Rating

Short answer: it depends on your rating, your time control and the site. In Lichess blitz, a typical game by a player rated 1200 scores 76%, and one by a player rated 2000 scores 83%. Chess.com uses its own formula, so its numbers differ. One game's accuracy swings a lot, so it says little about your level.

You finish a game, open the review, and it says 74%. Is that good? It feels like a school grade, so 74 feels like a C. It is not. Accuracy depends on your rating, the clock and the site that measured it, and one game's number moves around a lot.

We measured it. Below is the accuracy players actually reach at each rating, on both sites, from real games.

Typical accuracy by rating on Lichess

This is Lichess's accuracy, from 404,358 rated games played in September 2026 that someone had Lichess analyse. We computed it with Lichess's published formula and checked it against the figure Lichess shows on 202 player-games. Every one matched.

In each cell, the big number is the median: half of games score above it, half below. The small range under it is where the middle half of games land.

Lichess only has these evaluations for games someone asked it to analyse, about one game in ten. Lichess ratings also run on their own scale, usually higher than Chess.com's for the same player, so compare rows within one table, not across the two.

Typical accuracy by rating on Chess.com

This is Chess.com's own accuracy, from 21,112 rated games played in September 2026. Chess.com has no bulk download, so we sampled 3,000 players and every game of theirs that month.

Chess.com only reports accuracy for games someone ran Game Review on, so these are reviewed games, and players review some games more than others. A band marked "too few games" had fewer than 300 in our sample. Players rated 1200 to 1799 rarely turned up among the players we sampled; we are adding more.

Why Chess.com and Lichess give different numbers

Each site has its own formula, so the same game gets two different scores. Lichess publishes its method: it turns each position's engine evaluation into a winning chance, scores every move by how much chance it gave away, and blends two averages so that big mistakes count for more. Chess.com uses its own model.

So compare your number only with the table for the site that measured it. KingChess grades on its own scale too. Our free accuracy checker shows our grade and Lichess's side by side, so you can hold any game against the Lichess table above.

One game's accuracy says little

Look at the ranges, not just the medians. On Lichess, the middle half of blitz games by players rated 1200 score between 67% and 84%. For players rated 2000, it is 75% to 89%. Those ranges overlap a lot. A 1200 plays games that look like a 2000's, and the other way round.

Bullet is flatter still. On Lichess, the median bullet game scores 76% for players rated 1200 and 79% for players rated 2000. With seconds on the clock, everyone makes more loose moves.

Your average over many games tells you more than any single one. The moves behind the number tell you the most.

What to look at instead

Accuracy is an average, and an average can hide the one move that lost the game. At club level, games usually turn on a blunder: a piece left hanging, a fork walked into, a threat ignored. Count those instead. How many per game, and what kind keeps coming back.

Paste a game into our free accuracy checker and it marks every blunder and mistake. To test a single move you are unsure about, use the blunder check. If you want to read the numbers yourself, start with how the eval bar works.