How to beat the French Defense, by the numbers (808 top-player games)
Quick answer
What's the best way to beat the French Defense? Play 2.d4 d5 3.Nc3. It was the clearest result in the data, ahead from both sides on both sites: top players with White scored 72% in 110 games on Lichess (+5 vs their 67% White average) and 79% in 47 on Chess.com (+6 vs 73%, a hint), while French players facing it scored 66% in 73 on Chess.com (-1 vs their 67% Black average) and 59% in 63 on Lichess (-3 vs 62%). More broadly, an early d4 scored better for White than no-d4 set-ups: 78% in 95 vs 74% in 35 on Chess.com, 69% in 212 vs 63% in 73 on Lichess. Fast games, and different players chose different lines, so read it as a strong hint, not proof.
The French Defense (1.e4 e6) is the closed, counter-punching defence that a lot of 1.e4 players never learn to meet properly. We sorted every French in our two samples by what White actually played, move by move: 299 games among the top 50 Chess.com blitz players and 509 among the top 100 titled Lichess players. The French was about 11% of the 1.e4 games the Chess.com players had as Black, and 8% on Lichess. Score means points per game for the top player (win 1, draw ½). Compare with their averages (73% with White and 67% with Black on Chess.com, 67% and 62% on Lichess), not with 50%.
Short answer
- Overall, the French is close to par. Top Chess.com blitz players scored 77% in 130 games (4 points above their 73% average) with White against it; titled Lichess players 68% in 285 games (1 point above their 67% average). Playing the French themselves, they scored 70% in 176 games (3 points above their 67% average) on Chess.com and 62% in 239 games (right on their 62% average) on Lichess.
- Early d4 scored better for White, but it's not the whole story. With d4 on move 2 or 3, top players with White scored 78% in 95 games (5 points above the 73% average) on Chess.com and 69% in 212 games (2 points above the 67% average) on Lichess. Without it (King's Indian Attack, 2.Nf3 or 2.Nc3 lines): 74% in 35 games (1 point above the 73% average; under 50 games, a hint) and 63% in 73 games (4 points below the 67% average). But the French players' own results point the other way: they scored 70% in 156 games (3 points above the 67% average) on Chess.com and 65% in 171 games (3 points above the 62% average) on Lichess against an early d4, and 53% in 68 games (9 points below the 62% average) on Lichess against no-d4 set-ups. So d4 alone isn't the answer; which d4 line matters (see 3.Nc3 below). Compare the Caro-Kann, where delaying d4 scored better for White on both sites.
- 3.Nc3 was the best main move, from both sides. Counting every Black reply (Winawer, Classical, Rubinstein): White scored 72% in 110 games (5 points above the 67% average) on Lichess and 79% in 47 games (6 points above the 73% average; under 50 games, a hint) on Chess.com. French players facing it scored 66% in 73 games (1 point below the 67% average) on Chess.com and 59% in 63 games (3 points below the 62% average) on Lichess.
- The Tarrasch (3.Nd2) was the French players' favourite to face. Playing the French against 3.Nd2 they scored 80% in 25 on Chess.com (+13) and 82% in 25 on Lichess (+20). Small samples, so a hint, but a big gap on both sites.
- The Advance (3.e5) is popular, not deadly. White scored 73% in 15 on Chess.com (±0) and 62% in 25 on Lichess (-5); French players scored 69% in 32 (+2) and 61% in 35 (-1) against it. All hints (under 50 games).
1. Early d4 or not?
The biggest split in the data. “No d4” covers the King's Indian Attack (2.d3, or 2.Nf3 d5 3.d3), 2.Qe2, the Two Knights (2.Nf3/2.Nc3 d5 3.Nc3/3.Nf3), an early exd5 or e5 without d4, and other second moves. Transpositions count where they land: 2.Nc3 d5 3.d4 is a 3.Nc3 line, 2.Nd2 d5 3.d4 is a Tarrasch. Rows under 50 games are hints.
| White's approach | Chess.com top 50 with White | Chess.com top 50 playing the French | Lichess titled with White | Lichess titled playing the French |
|---|---|---|---|---|
| d4 on move 2 or 3 | 78% in 95 (+5) | 70% in 156 (+3) | 69% in 212 (+2) | 65% in 171 (+3) |
| no d4 in White's first 3 moves | 74% in 35 (+1), hint | 68% in 20 (+1), hint | 63% in 73 (-4) | 53% in 68 (-9) |
| 3.Nc3 (all Black replies) | 79% in 47 (+6), hint | 66% in 73 (-1) | 72% in 110 (+5) | 59% in 63 (-3) |
2. Every White system against the French
Sorted by how often it came up. The 3.Nc3 lines are split by Black's reply (3...Bb4 Winawer, 3...Nf6 Classical, 3...dxe4 Rubinstein, which also counts 3.Nd2 dxe4). Each cell is the top player's score and the gap to their average; under 50 games is a hint, under 8 is too few to read.
| White's system vs 1...e6 | Chess.com top 50 with White (avg 73%) | Chess.com top 50 playing the French (avg 67%) | Lichess titled with White (avg 67%) | Lichess titled playing the French (avg 62%) |
|---|---|---|---|---|
| Classical (3.Nc3 Nf6) | 72% in 20 (-1), hint | 73% in 31 (+6), hint | 67% in 63 (±0) | 48% in 33 (-14), hint |
| Advance (3.e5) | 73% in 15 (±0), hint | 69% in 32 (+2), hint | 62% in 25 (-5), hint | 61% in 35 (-1), hint |
| Tarrasch (3.Nd2) | 81% in 13 (+8), hint | 80% in 25 (+13), hint | 64% in 40 (-3), hint | 82% in 25 (+20), hint |
| Exchange (3.exd5) | 85% in 13 (+12), hint | 71% in 21 (+4), hint | 71% in 21 (+4), hint | 70% in 32 (+8), hint |
| King's Indian Attack / d3 set-ups (incl. 2.Qe2) | 69% in 13 (-4), hint | 75% in 12 (+8), hint | 62% in 30 (-5), hint | 50% in 16 (-12), hint |
| Rubinstein (3.Nc3/3.Nd2 dxe4) | 90% in 15 (+17), hint | 6 games, too few | 77% in 24 (+10), hint | 69% in 26 (+7), hint |
| Winawer (3.Nc3 Bb4) | 82% in 14 (+9), hint | 53% in 33 (-14), hint | 88% in 12 (+21), hint | 5 games, too few |
| Early exd5 without d4 (2.Nf3/2.Nc3 d5 3.exd5) | 61% in 9 (-12), hint | – | 58% in 18 (-9), hint | 56% in 16 (-6), hint |
| Two Knights (2.Nf3/2.Nc3 d5 3.Nc3/3.Nf3) | 7 games, too few | 3 games, too few | 73% in 15 (+6), hint | 54% in 13 (-8), hint |
| 3.Nc3, other Black 3rd moves | 1 game, too few | 3 games, too few | 73% in 15 (+6), hint | 5 games, too few |
| 2.d4 without 2...d5 | 4 games, too few | 3 games, too few | 7 games, too few | 6 games, too few |
3. How to choose your anti-French
- You want the best-scoring main line: 2.d4 d5 3.Nc3. It asks the French player to know the Winawer or the Classical, and their scores dipped below average against it on both sites.
- You like space and a clear plan: the Advance (3.e5) is fine, but French players are ready for it: White scored at or below average with it on both sites here (hints).
- You want low theory: the King's Indian Attack (2.d3 and Nd2, g3, Bg2) avoids the French player's preparation, but it scored 69% in 13 on Chess.com (-4) and 62% in 30 on Lichess (-5) for White here (hints), and French players' own results against it were mixed. Easier to play, not more effective.
- Think twice about the Tarrasch (3.Nd2) in blitz: it's the line French players scored best against in both samples.
- Full plans and a line to play are in the French Defense guide.
4. What matters more than the line
- Your opponent's French. French players usually have one favourite answer to 3.Nc3 and one to the Advance, and they repeat it. Check before the game: the free opponent preview shows any Chess.com or Lichess player's most-played openings with each colour, and the $5 report shows the exact lines where they score worst.
More from the same data: how to beat the Caro-Kann, the best chess openings for White, the best chess openings for Black, the best Sicilian variation and the best response to the London System.
Method: Chess.com sample is the top 50 of the Chess.com live blitz leaderboard on 2026-10-20, each player's most recent rated standard blitz games (up to 150, at most six monthly archives) from the public Chess.com Published-Data API, games played 2026-03-21 to 2026-10-20. Lichess sample is the last 150 rated games of the top 100 titled players on the Lichess leaderboards (2026-10-18). French Defense = 1.e4 e6; White's system is read from the moves, not the opening name. A game between two sampled players counts once per sampled player. Rows under 50 games are hints; headline claims need 50+ games except where marked. OpeningSpy is independent and not affiliated with Chess.com or Lichess.