I agree with M.Ayton and add that in first game 18.Rb4!? Qc5 good move for White is not the direct 19.Ne4 where Black can have counterplay with 19...Qg1 but in fact 19.Qe2 and no hope for Black to play Qg1 because Qf2 leads to advantage to White in the queenless ending.

After 18...Bc6! 

A) 19.Ne2 Bb5! 20.a4 Qc5 the novelty of the game Hort-Nogueiras,Rotterdam 1989 ; 20.Qb3 Na5! (20...Qc5) 21.Qc3! Nc6 22.Rb1 d4= ; 20.Qf3 d4 21.g4 Nce3 22.gxf5 Nxc2+ 23.Kd2 Ne3 24.Ke1 Be2 25.Kxe2 Qc2+ 26.Bd2 Qd1= ; 20.Nd4 Nxe5 Scholz-Hannelbauer, corr 1988

B) 19.h6 Rg6! 20.h7 Rh8 21.Ne2 Rg7 22.g4 a5!? or 22.c3 Bb5 23.Qf3 Rgxh7 or 22.Nd4 Nxd4 23.Qxd4 Rhxh7=  ; 21.Qd1 Rg7! Chigvintsev-Pokrasenko,Novosibirsk 2002

C) 19.g3 Qa5 20.h6 Rh6 21.h7 Rh8 22.Nxd5 Bxd5 23.Qc3 Qc5 24.g4 Rxg4 25.Bxc4 Qg1+ 26.Bf1 Kb8 27.Be3 Nxe3 28.Qxe3 Qxe3 29.Rxe3 a5 30.Rd4 Rxh7

After 7 ..o-o the most critical line is certainly 8 Bd3 - and then if 8.. Nc6 9 Qh5 instead of 9 Nf3. After 8 .. f5 white also tends to delay Nf3 a few moves in favour of ef/Bg5/Qh5 etc. I can't see this changing really: Bd3 first just works better for white.

One symptom of this is things like 8 Nf3 Nc6 9 Bd3 f5 10 ef Rxf6 11 Bg5 e5!? working quite a bit better than the equivalent lines without Nf3 & Nc6 included. 

The various French books seem to be reasonably kind about the status of 7 .. Nf5 although hardly anyone seems to play it. 7.. Kf8/6 .. Qc7/Qa5 etc aren't so awful either.

In general there are an awful lot of at least nearly playable solutions for black, although how many (any?) of them are truly totally watertight is perhaps questionable.

I wonder that too: after 7.Qg4 0-0, what is the "most respected" from the theory, 8.Bd3 or 8.Nf3? And why?

ArKheiN wrote on 05/11/08 at 12:24:36:


Poisoned pawn variation , it is quite risky, especially if white is prepared you have to play a lost endgame and hoping white to go astray. 3...Nc6 is quite interesting and seems to be a healthy alternative. I recommend two variations against Winaver after 7.Qg4 0-0 and if Bd3 then Qa5 a rather understimate variation from the theory.And the other is after 7.Qg4 Nf5!? better than it's reputation.Khalifman's book on Winaver says it is bad, but after checking the lines with Rybka, I found some improvements in the refutation line.
Also if you want to suprise even 3...h6!? is good both against Winaver or even Tarrasch

Pingudon wrote on 05/07/08 at 14:36:43:

Poisoned pawn for me with both colours. Thing is that you must focus on total annihilation of the white center, while keeping an eye on those nasty pawns on the kingside and moves like Nb5.

If you are playing against the Winawer, you are playing White. But you seem to be asking what to play as Black against 3.Nc3. Actually I would recommend having more than one option available, since 3.Nc3 is so popular and critical. 

I'm not sure the gxf6-Burn is that risky; at GM level it has a relatively high drawing rate. But you have to be comfortable with an unbalanced position (bishop pair with a fractured pawn structure). Of course, the Burn only covers 3...Nf6 4.Bg5, you also need a line against 4.e5. 

The Winawer Poisoned Pawn should be well playable and a lot of fun on almost any level, since it is so easy for White to go wrong (as both Shirov and Cheparinov have recently done!). But White has a lot of playable alternatives on move 4, 5 and 7, so building a repertoire against all of them can take some time. Check out Chesspublishing.com for both the gxf6-Burn and the Winawer, and also Uhlmann's "Winning with the French" for the latter. For the Winawer sidelines, I suggest you use either "Play the French 3" by Watson or "French Defence 3.Nc3 Bb4" by Psakhis as a starting point and pick one good answer to each line. Good luck!