The following position shows that Black tries to enter the Tartakower variation (4.. Nf6). However, White avoids it by playing 5. Ng3. In this case, I'm lost about what to do as Black. I also fear White's bishop pair which points to my kingside.

rnbqkbnr/pp2pppp/2p5/3p4/3PP3/2N5/PPP2PPP/R1BQKBNR b KQkq - 0 1

3... dxe4 4.Nxe4 Nf6 5.Ng3

The reason I avoid the main line 4.. Bf5 and plays 4.. Nf6 is that in the 4.. Bf5 variation, White often expands on the kingside and then exchange the light-squared Bishop. After this, White could have expanded a lot on the kingside and Black ends up with weak light squares on the kingside.

With the Tartakower, on the other hand, I often get a comfortable position. The only problem now is that when White plays 5. Ng3 and tries to transpose to the 4.. Bf5 variation, I don't know anymore what to do.

Stockfish 15 suggests 5.. h5 after 5. Ng3. However, it's hard to see the plan involving 5.. h5 and as a human, in my eye, it becomes hard to castle kingside after playing 5.. h4.

Another problem after 5. Ng3 is how to develop the light-squared Bishop. For example, playing e6 later on blocks the light-squared Bishop. Now, I can probably do Bd6 followed by Bc6 but the problem is the tempo. That is, if I get behind in development as I also need to develop my other Knight which will be blocked by Bd6, White could be already starting their attack on the Kingside.

1 Answer 1


Summary (strong ideas for Black)


According to the analysis of IM Sielecki and GM L'Ami:

  • play ..h5, threatening to dislodge the knight
  • develop the bishop to g4 (provoking the weakening f3, gaining a tempo on the queen or pinning a knight)
  • quickly strike at the center with a timely ..c5
  • develop the knight to Nc6 (after striking with ..c5)

..with easy play. Don't fear White attacks too much. In this line, it is Black who is more likely to generate initiative. Also, don't worry too much about castling - and you can also consider castling queenside (as recommended by the engine checking some lines from the mentioned repertoires).


According to IM Banzea:

  • play g6, restricting the knight
  • fianchetto the bishop to g7
  • attack the center with ..c5 to open the bishop's diagonal immediately after the fianchetto
  • take on d4, if White doesn't play dxc5
  • develop the knight not to c6 (!), but to d7 and bring it to c5 or win back the pawn on c5 if White takes dxc5

IM Banzea considers the good alternatives; in his taste: ..h5 ("best, but strange" if White plays f3), ..c5 ("too concrete"), ..e6 ("tricky due to passive bishop"), ..e5 ("dull"); he considers his recommendation ..g6 to be "unusual" for the Caro-Kann due to the fianchetto, but simpler than the alternatives.

Long version:

5. Ng3 isn't a good try by White to play for an advantage. On the contrary, the engine already likes Black's position. It's almost never seen on higher levels (i.e. in the master's database). On amateur level however, it is frequently played by (up to ~20%! of) players who have semi-studied their line against the Caro-Kann and are out of theory after move 4, disregarding Black's move entirely and reacting in the same way as after 4. ..Bf5 (maybe hoping to transpose even though they're not regaining the tempo by hitting the bishop). Interestingly, players with lower lichess ratings who are less likely to study theory are also less likely to play that move!


I'd like to present 5. ..h5 (a very strong and aggressive move threatening to dislodge the knight), as it is explained / recommended in both IM Sielecki's Keep it Simple for Black and Erwin L'Ami's LTR Caro-Kann repertoires (although L'Ami also covers 4. ..Bf5).

Both claim, that play is very easy for Black, with easy development and even initiative. If White doesn't prevent ..h4, we'll play ..h4 (kicking the knight), often followed by developing the bishop to g4 which often tempts White players to weaken their structure by playing f3.

If White prevents this idea by playing 6. h4, Black can play Bg4, but even more direct is the recommended strike at the center and prepare to develop the knight to c6: 6. ..c5!?.

White can take the pawn, which leads to a situation comparable to the QGA. Black will win the pawn back (not immediately, but eventually), starting with trading queens and playing 8. ..e6 (attacking the pawn with the bishop). If White tries to hold it by means of playing b4, Black plays ..a5 and starts to rupture the queenside. It is worth looking at this line, even if it's not attractive for White objectively, Black should know how to get the pawn back with interest. White will end up with an isolated a-pawn that will be a liability rather than a strength (according to IM Sielecki).

Here is a sample line from the repertoire (not in it's entire length):

 [FEN ""]
 1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nf6 5. Ng3 h5 6. h4 c5 7. Nf3 Nc6 {L'Ami gives ..Bg4} 8. Be3 cxd4 {here, Sielecki states that 8. ..Bg4 is possible, playing with a pawn down for some time, but considers his recommendation simpler} 9. Nxd4 Bg4


EDIT: I just checked Banzea's Chessable Repertoire The Caro-Kann simplified and found that he considers ..g6 to be the simplest choice for Black. See this line in action in his YouTube video with Caro-Kann middlegame tips for beginners, as well as in his Caro-Kann speedrun (2000 Elo).

See the summary above for a description of ideas in this line for Black.

Sample line (not entire length):

[FEN ""]
1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nf6 5. Ng3 g6 6. Nf3 Bg7 7. Bc4 c5

Just like in the line above, White cannot keep the pawn if he takes dxc5. You should trade the queens in this case and play Nbd7 in order to win back the pawn on c5. If White tries to defend the pawn with Be3, use your knight on f6 to attack that bishop and remove it as a defender. White cannot play b4 to keep the pawn immediately, as this loses due to the opened long diagonal.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.