Sign in

St. George Defense (1. )

The St. George Defense is 1. . The a6 pawn prepares : it protects b5 if a bishop hits it, and it clears b7 for your bishop, which then aims at e4 down the long diagonal. Between 900 and 1700 you score 48% with it (N = 53,953 Lichess games, September 2026). It shows up in about 2 games in 1000, and this page follows the line that players rated over 2200 choose against it, move by move.

Lichess games from September 2026 (30,577,162 rated 900 to 1700) and 31,791 over-the-board games rated 2300+ from the last 12 months. Data updated: October 14, 2026.

At a glance

Moves
1.
ECO
B00
Amateur games
53,953
Of all games rated 900 to 1700
0.2%
White's score
Lichess 900–1700: 52.0% · Online, 2200+: 50.6% · Over the board, 2300+: 66.7%
Evaluation after
+0.49

4 named lines in this family; 1 with their own page.

By rating

The opening's share in each Lichess rating band.

What amateurs play and what strong players play

After , 37 in 100 opponents at your level play 2., 33 bring out the knight with 2. and 11 put the bishop on c4 with 2.. Among players rated over 2200 online, 2. climbs to 75 in 100. Against 2., close to half of the players in both groups answer . Next comes at your level, 17 in 100, and among strong players, 21. After the two groups split: at your level 40 in 100 play 3. and only 4 go for 3., while strong players prefer 3., 33 in 100, or 3., 22. After 3., 94 in 100 players at your level follow up with , aiming the bishop at e4.

After 1. · N = 56,081
MoveLichess 900–1700White's scoreOnline, 2200+Over the board, 2300+
36.6%52.7%75.3%100.0%
33.1%52.4%9.7%no data
11.4%51.6%1.9%no data
5.9%50.9%4.4%no data
3.2%54.4%no datano data
2.5%48.6%no datano data
no datano data2.5%no data
no datano data2.0%no data
After 1. 2. · N = 23,603
MoveLichess 900–1700White's scoreOnline, 2200+Over the board, 2300+
46.5%50.7%46.4%no data
17.4%53.6%6.2%no data
10.6%53.3%21.4%no data
5.7%53.4%no datano data
4.7%54.4%7.6%no data
4.6%55.8%no datano data
no datano data4.7%no data
no datano data3.9%no data
After 1. 2. · N = 11,474
MoveLichess 900–1700White's scoreOnline, 2200+Over the board, 2300+
40.1%51.7%17.2%no data
13.9%48.1%10.4%no data
10.0%52.4%no datano data
7.3%49.2%6.8%no data
5.0%50.3%3.3%no data
3.7%54.0%33.3%no data
no datano data21.8%no data
After 1. 2. 3. · N = 373
MoveLichess 900–1700White's scoreOnline, 2200+Over the board, 2300+
93.6%52.1%100.0%no data
6.4%58.3%no datano data
After 1. 2. 3. · N = 234
MoveLichess 900–1700White's scoreOnline, 2200+Over the board, 2300+
48.7%54.4%58.3%no data
31.2%60.3%no datano data
11.1%53.8%6.7%no data
9.0%81.0%3.9%no data
no datano data17.3%no data
no datano data5.9%no data
no datano data3.5%no data

What to do here

The strong-player line is 2. 3. 4. 5. . Everything revolves around the e4 pawn. Your opponent takes the center with two pawns and you don't contest those squares with your own. You hit e4 from a distance instead: the bishop from b7, down the diagonal through c6 and d5, and the knight from f6. By the end of the line your f8 bishop has a way out, the last step before you can castle.

The move order has a reason. comes before because the f1 bishop already sees b5 through e2, d3 and c4; with a pawn on a6, if it takes on b5 you take back. clears b7, and lands there with an attack on e4. That's why strong players go 3.: the bishop guards e4 and also hits b5, which your a6 pawn still covers.

4. develops but doesn't protect e4. takes advantage: now two of your pieces attack e4 and only the d3 bishop defends it. 5. adds a second defender. With two against two, taking on e4 would cost you the knight for a pawn. In that position 82 in 100 strong players choose 5., but only 20 in 100 opponents at your level do. When it doesn't come, count the defenders of e4 before anything else.

opens the diagonal toward c5 and b4 for your f8 bishop, and once it's out you can castle short. It also covers d5: if your opponent pushes the e4 pawn to e5 to kick your f6 knight, the knight has d5, guarded by the e6 pawn and the b7 bishop.

3., the strong players' second choice at 22 in 100, attacks b5 with a pawn. If you push on with , your pawn takes c3 away from the b1 knight, but nothing defends it until you move the e7 pawn and your f8 bishop sees b4. At your level this move hardly ever appears: fewer than 4 in 100 opponents choose it.

At your level the usual reply to is 3., and 4. reaches the Polish Variation: your opponent brings out the knight and bishop in a different order, where you score 47.3 out of 100. Against 3., picked by 14 in 100, the pawn hits b5 and you score 51.9 out of 100. Against 2., be careful: the bishop is eyeing f7, and is the costliest mistake on this page, because f7 has only your king to defend it.

The strong players' line

The most repeated move in games rated 2200+ online and 2300+ over the board, while at least 20 games play it
MoveStrong gamesChosen by
1,66975.4%
1,65346.4%
82533.3%
270100.0%
25458.3%
34766.0%
17581.7%
159100.0%

The mistakes that cost the most games

Blunder

After 2., only your king defends f7, and does nothing about it. Then 3. 4. 5. : the queen gives check and hits your a8 rook through the empty b7 square. You lose castling, and the rook for the bishop. Play . 45 in 100 play it; it often goes unpunished: 49.9 against 52.

45 in 100 play it in this position. It gives away 3.3 pawns of advantage. It scores 49.9%; scores 52.0%.

Inaccuracy

Inaccuracy, unpunished at your level. is the strong players' move, but 3. attacks the pawn and you push on with . That pawn takes c3 from the knight, which goes 4., and after 5. it sits far forward, guarded only by your f8 bishop. is better, though you score 49.3 with , against 46.7.

47 in 100 play it in this position. It gives away 0.6 pawns of advantage. It scores 49.3%; scores 46.7%.

Inaccuracy

Inaccuracy, rarely punished at your level. After 3., develops nothing and ignores the center. Then 4. 5. 6.: the a4 pawn and the d3 bishop hit b5, which only your a6 pawn covers, so your b7 pawn can't reach b5 without being lost. Play . 26 in 100 play it; you score 47.8, against 49.

26 in 100 play it in this position. It gives away 0.6 pawns of advantage. It scores 47.8%; scores 49.0%.

Inaccuracy

Inaccuracy that barely gets punished at your level. attacks e4, but after 3. your queen is in the middle of the board too early. Then 4. 5.: the pawn hits the queen and sends it back home, facing pawns on c4 and d4. is better. 6 in 100 play it; you score 46.6, against 46.7.

6 in 100 play it in this position. It gives away 0.6 pawns of advantage. It scores 46.6%; scores 46.7%.

Mistake

After 2., leaves f7 guarded only by your king and empties b7. Then 3. 4. 5. : the pawn blocks the check and attacks the queen, but from d5 it sees your a8 rook and grabs it. You lose castling and the exchange. Play . 4 in 100 play it; you score 40.2, against 52.

4 in 100 play it in this position. It gives away 2.6 pawns of advantage. It scores 40.2%; scores 52.0%.

Inaccuracy

Inaccuracy, unpunished at your level. After 3., doesn't challenge the pawns on c4, d4 and e4. Then 4. 5. 6. : you give up your bishop for the knight and the c3 pawn shores up d4. Play , which hits both c4 and e4. 27 in 100 play it; you score 49.6, against 47.2.

27 in 100 play it in this position. It gives away 0.9 pawns of advantage. It scores 49.6%; scores 47.2%.

How players castle

Lichess 900–1700 · the 6 most frequent combinations
WhiteBlackGames
18,686
Not castled15,165
Not castledNot castled8,164
3,004
Not castled2,777
Not castled2,629
Over the board, 2300+ · the most frequent combination
WhiteBlackGames
3

Variations

Other named lines

Polish Variation

1. 2. 3. 4. · B00

1,705 amateur games · White's score: 52.7%

Zilbermints Gambit

1. 2. · B00

358 amateur games · White's score: 53.5%

San Jorge Variation

1. 2. 3. 4. 5. 6. · B00

62 amateur games · White's score: 52.4%

Lichess games from September 2026 (30,577,162 rated 900 to 1700) and 31,791 over-the-board games rated 2300+ from the last 12 months.

Data from Lichess (CC0) and Stockfish 19. Text written with AI from this data and checked automatically against it.

Data updated: October 14, 2026.

How we make the data