Skip to content
TrueEloPlay what wins.
Score a position

Three Knights Opening

5 variations · 14M games from Lichess · Updated January 2026

8
7
6
5
4
3
2
a1
b
c
d
e
f
g
h
3...Bb4
Analyze this line move by move

White expected score · Overall

51.6%

45%55%
Floor51.5%

the worst score you would still get, 90% confident. Its best is the same figure — the band is too tight to plot.

Measured over 14M games at this level and clock.

Playablewins about as often as it loses at this level

Roughly even — the middlegame decides more than this opening choice does.

Outcomes for White

Draw 4.0 · Score = win + ½ of the draws · 14M games

Rank among the families White can choose here
#32 of 56
Games in this context
14M
Best context
60.2% at 2200-2499 Rapid
Worst context
46.5% at 0-999 UltraBullet
Variations measured
5
Average game length
33 moves

Best and worst are the strongest and weakest squares with at least 500 games behind them. The table further down shows every square, including thinner ones.

What to play from here

White · Overall
#1Three Knights Opening: Steinitz-Rosenthal VariationBest here
1.e4 e5 2.Nf3 Nc6 3.Nc3 g6 4.d4 exd4 5.Nd5
Floor 54.4% · 1.4K games
56.5%
#2Three Knights Opening
1.e4 e5 2.Nf3 Nc6 3.Nc3
Floor 51.6% · 13M games
51.6%

These are the moves players actually choose from here, ranked by how they score. A higher score is a better bet on the numbers alone — familiarity and playstyle still matter.

Where it holds · White score by rating × clock

Green is clearly ahead for White, grey is around even, red is behind. Tap a cell to measure there, and again to clear. It gets better as the rating rises: 48.8% at 0-999, 58.0% at 2500+.

RatingUBBulleBlitzRapidClass
0-999
1000-1199
1200-1399
1400-1599
1600-1799
1800-1999
2000-2199
2200-2499
2500+

Expected score matrix by rating bracket and time control

  • 0-999 rating, UltraBullet: 46.5% expected score (1,872 games)
  • 0-999 rating, Bullet: 48.9% expected score (314,481 games)
  • 0-999 rating, Blitz: 50.1% expected score (681,281 games)
  • 0-999 rating, Rapid: 50.6% expected score (419,206 games)
  • 0-999 rating, Classical: 47.9% expected score (3,469 games)
  • 1000-1199 rating, UltraBullet: 49.1% expected score (3,354 games)
  • 1000-1199 rating, Bullet: 50.9% expected score (515,799 games)
  • 1000-1199 rating, Blitz: 51.6% expected score (1,029,865 games)
  • 1000-1199 rating, Rapid: 51.8% expected score (565,437 games)
  • 1000-1199 rating, Classical: 50.4% expected score (12,008 games)
  • 1200-1399 rating, UltraBullet: 51.0% expected score (4,385 games)
  • 1200-1399 rating, Bullet: 51.2% expected score (671,222 games)
  • 1200-1399 rating, Blitz: 51.6% expected score (1,272,017 games)
  • 1200-1399 rating, Rapid: 51.8% expected score (647,934 games)
  • 1200-1399 rating, Classical: 50.6% expected score (25,090 games)
  • 1400-1599 rating, UltraBullet: 52.8% expected score (4,747 games)
  • 1400-1599 rating, Bullet: 51.1% expected score (748,926 games)
  • 1400-1599 rating, Blitz: 51.5% expected score (1,363,917 games)
  • 1400-1599 rating, Rapid: 50.8% expected score (580,204 games)
  • 1400-1599 rating, Classical: 49.4% expected score (34,203 games)
  • 1600-1799 rating, UltraBullet: 55.4% expected score (2,898 games)
  • 1600-1799 rating, Bullet: 51.7% expected score (730,904 games)
  • 1600-1799 rating, Blitz: 52.0% expected score (1,112,734 games)
  • 1600-1799 rating, Rapid: 52.6% expected score (385,476 games)
  • 1600-1799 rating, Classical: 52.9% expected score (19,011 games)
  • 1800-1999 rating, UltraBullet: 54.2% expected score (1,912 games)
  • 1800-1999 rating, Bullet: 51.6% expected score (619,192 games)
  • 1800-1999 rating, Blitz: 52.4% expected score (668,868 games)
  • 1800-1999 rating, Rapid: 53.4% expected score (166,770 games)
  • 1800-1999 rating, Classical: 56.1% expected score (5,663 games)
  • 2000-2199 rating, UltraBullet: 60.0% expected score (724 games)
  • 2000-2199 rating, Bullet: 51.7% expected score (394,147 games)
  • 2000-2199 rating, Blitz: 54.3% expected score (230,921 games)
  • 2000-2199 rating, Rapid: 57.0% expected score (38,025 games)
  • 2000-2199 rating, Classical: 57.7% expected score (776 games)
  • 2200-2499 rating, UltraBullet: 54.4% expected score (274 games)
  • 2200-2499 rating, Bullet: 52.8% expected score (206,576 games)
  • 2200-2499 rating, Blitz: 57.4% expected score (51,830 games)
  • 2200-2499 rating, Rapid: 60.2% expected score (5,984 games)
  • 2200-2499 rating, Classical: 60.9% expected score (86 games)
  • 2500+ rating, UltraBullet: 60.0% expected score (64 games)
  • 2500+ rating, Bullet: 54.5% expected score (33,571 games)
  • 2500+ rating, Blitz: 58.7% expected score (2,578 games)
  • 2500+ rating, Rapid: 67.0% expected score (137 games)
  • 2500+ rating, Classical: 50.0% expected score (3 games)

5 variations · ranked by floor

Floor — the worst score you would still get, 90% confident.

1
Three Knights Opening: Steinitz-Rosenthal Variation1.4K games
56.5%Floor 54.4%
2
Three Knights Opening: Steinitz Defense195K games
52.4%Floor 52.3%
3
Three Knights Opening13M games
51.6%Floor 51.6%
Three Knights Opening: Schlechter Variation75K games
50.4%Floor 50.1%
Three Knights Opening: Winawer Defense519K games
49.6%Floor 49.5%
Analyze Three Knights Opening PositionsSee what each move is worth at your rating, move by move

Methodology

Across every rating and clock, the Three Knights Opening family scores 51.6% for White — we are 90% confident the real figure is between 51.5% and 51.6% — from 14M Lichess games across 5 variations. Win rate, draw rate and loss rate come from games played December 2023 to November 2025.

How the figures are made
Whose games these are
14M+ rated games played on Lichess (2023-12 to 2025-11). The games are Lichess’s and free for anyone to use. What is ours is the reading of them. Last rebuilt January 2026.
Which games we count
  • Rated games only — nothing casual.
  • Both players within 200 rating points of each other. A mismatch tells you about the gap, not about the opening.
  • At least 2 moves played.
  • A game is filed under White’s rating.
Why every number comes with a range
A number off a few hundred games is a guess. The same number off millions is a fact. The range is what tells them apart, and it is what the lists rank on — so an opening cannot reach the top on a lucky run of forty games. The more games behind a figure, the narrower its range.
How much we need before we say anything
An opening page needs 100 games. A gambit needs 50 games at a rating before we will say whether it pays there. The chaos and trap lists need 500 games overall and 250 games at one rating or clock.

Frequently Asked Questions

What is the Three Knights Opening win rate?

Across the Three Knights Opening's 5 variations, White wins 49.5%, draws 4.0% and loses 46.4%. Counting a draw as half a point, that is an Expected Score of 51.6%. It rests on 14M rated games from Lichess.

How many Three Knights Opening variations are there?

5 Three Knights Opening variations have a page of their own, each broken down by rating and by clock.

At what rating does the Three Knights Opening perform best?

The Three Knights Opening family performs best for White at 2500+, with an Expected Score of 54.8% (36K games) and the highest Floor of any rating with enough games to read. Performance may vary by specific variation.

Is the Three Knights Opening better in blitz or rapid?

For White, the Three Knights Opening scores about the same in both: an Expected Score of 51.7% in Blitz (6.4M games) against 51.7% in Rapid (2.8M games).

How sharp is the Three Knights Opening?

Three Knights Opening games are 96.0% decisive (non-draws), with an average length of 33 moves. Sharpness varies by specific variation.

Is the Three Knights Opening a good opening family?

The Three Knights Opening family is about level overall. Across its games, White wins 49.5% and loses 46.4%. Results differ between its 5 variations, so open the one you play for its own figures.