[FEN ""]

1.e4 e5 2.f4 exf4 3.Nf3 c5

What's the refutation of Black's unusual third move, or anyway the best way to play against it? Which is stronger, the Open Variation 4. d4 cxd4 5. Bxf4, or the Closed Variation 4. Bc4 Nc6 5. d3?

If there is no refutation (as seems to be the case), then why is 3...c5 hardly ever played, or mentioned in the books? Is it because there are so many good lines against the King's Gambit that nobody needed another one? Is it that the people who like to play c5 usually play it on the first move?

5 Answers 5


My initial reaction is that 3...c5 is very weak just by looking at it, which is why it is so rare. The ECO key that comes with ChessBase's Mega databases is very detailed, but even this does not even have its own specific sub-key.

There are 112 games in Mega 2020 database with 3...c5. The most popular response, played in 87 of the 112 games was 4.Bc4, which had a winning percentage of 64.9%, followed by 4.d4, which was played in 17 games and had a winning percentage of 61.8%). 4.d3 (6 games/33.3%) and 4.Nc3 (2 games/50.0%) were also played.

Obviously, 4.Bc4 and 4.d4 are the choices, but you also have to think about what to play stylistically: If you play the King's Gambit, you like open, tactical, positions, so it does not make sense to play the latter two moves at all unless they eventually become the natural move to finish development later. You still want to strive for the quick development of the kingside, and play down the f-file and open center.

3...c5 is weak, but as I looked through the games, the only plan that white needs to fear is the Be7 and g5 plan (possibly g5 first), holding the pawn. Black scored in the 40-50% range in most of those lines. That might bode well for the 4.d4, and early Bxf4 plan by white.

Now that I have discussed what humans play, I am going to see what Stockfish 11 says. After looking at many lines, a lot of which are below, but that is only a small fraction of what I actually looked at, I cannot say that Stockfish found any pure refutation, but I suspect that the you will do well following the main line because if black plays 3...c5, it is because black is a weak player to start with...deny black holding f4, and you will probably win most of the time as the human stats above show.

4.Bc4 and 4.d4 early on both come in with the same evals, but as the depth gets deeper (depth 33), 4.Bc4 is preferred by Stockfish, but they are both the typical minor pluses that computers typically give white in the opening, about +.5, but as I force in lines, and go deeper, everything levels out a lot. Here are some sample lines.

One last note about the play: The reason that white can play the early Bc4 in these lines, and not worry about g5, it that it seems tactically refuted if black tries to save it. White is playing a gambit line, so I was not really worried about trying to get the pawn back simply because we are down a pawn, but the computer might be; nevertheless, I was concerned because it gets in the way of developing the Bc1 and opening the f-file, so I did lean toward recovery.

Please note that there are also extensive comments within these lines.

 [FEN ""]

 1. e4 e5 2. f4 exf4 3. Nf3 c5 4. Bc4 Nc6 5. d3 {This is Stockfish's favorite, and is indicative of trying to stop the early g5 plan. It still is not the most human move, which I will also cover.} (5. d4 cxd4 6. Bxf4 Nf6 7. O-O Nxe4 8. Bd5 $1 {This leads to mind-boggling complications.} (8. Re1 d5 $17) 8... Nf6 9. Ng5 Nxd5 10. Nxf7 Qf6 (10... Kxf7 11. Qh5+ Ke6 12. Nc3 {And the other rook comes in and wins...+8.}) 11. Bg5 Qg6 12. Qe1+ Be7 13. Bxe7 Ne3 14. Rf2 Nxe7 15. Nxh8) (5. O-O $6 {This is slow, but leads to typical positions.} g5 6. d4 cxd4 7. Bxf7+ (7. g3 $5) 7... Kxf7 8. Nxg5+ Qxg5 9. Bxf4 Qg6 10. Be5+ Ke8 11. Bxh8 b6 {With wild play that the computer likes barely for black, but I do not believe that a human will survive.}) 5... Bd6 $5 (5... g5 $2 6. h4 g4 7. Ng5 $18) (5... Be7 6. Bxf4 Nf6 7. Nc3 d6 8. O-O O-O 9. Qd2 $16) 6. O-O Nf6 (6...Nge7 $4 7. Ng5 O-O 8. Qh5 $18) 7. Nc3 (7. Qe1 $5) (7. e5 $2 {This does not work due to tactics and threats of smothered mate, and the Qd1 and Bc1 hanging if the queens get traded.} Nxe5 8. Re1 O-O 9. Nxe5 Bxe5 10. Rxe5 d5 11. Bb3 c4 12. dxc4 Ng4 13. Re2 dxc4 14. Qxd8 Rxd8 15. Nc3 cxb3 16. axb3 g5 $17) 7... O-O 8. Bd2 a6 9. a4 {And the computer thinks this is roughly equal, but white is more fun to play.}
  • Wow, thanks. What's the evaluation (yours and 11's) of that line ending with 15 Nxh8?
    – bof
    Commented Feb 22, 2020 at 14:45
  • 1
    Believe it or not, the comp still thinks that is equal. Commented Feb 22, 2020 at 14:51
  • 1
    @bof One thing to note in that line, is that black is on a razor's edge. There were MANY more alternatives, and they lost. Even at the end, Qb6 is the only move. So follow the general advice, and look for tactics along the way. Commented Feb 22, 2020 at 15:25
  • I'm sure I would never have thought of 9.Ng5 over the board. Is that a "computer move" or is it a tactic that would come naturally to a human expert/master?
    – bof
    Commented Feb 23, 2020 at 1:45
  • @bof I agree, it is not super intuitive, but a lot of opening prep today is computer-driven. If you find that with a comp, a master could possibly find the rest over the board. Commented Feb 23, 2020 at 1:49

White's f4 sacrifices a pawn for rapid development and the possibility of an attack against the uncastled black king. c5 as a response from black addresses none of the issues in his position and just wastes a move making the gambit all the more effective. White's plan remains on track - Bc4, O-O, win the pawn back on f5 and smash through on f7.

  • No it makes sense, look at my comment Commented Feb 22, 2020 at 12:09

It does make sense because even the computer says black is better ( -0.5 ) after full development and entering in the middlegame.

3...c5 is a very good move through because it takes d4 square for your pawn. 4. c3 is stupid because you must continue with development and attacking. It gives black an equal or even better position.

Look at this line:

[FEN ""]   
 1. e4 e5 2. f4 exf4 3. Nf3 c5 4. Bc4 Nc6
    5. d4 cxd4 6. Bxf4 Nf6 7. O-O Nxe4 8. Ne5 Nxe5
    9. Bxe5 d5 10. Qh5 g6 11. Qf3 f6 12. Bxd4 Bf5
    13. Bb3 Bb4 14. c3 Ba5 15. Re1 Qd6 16. g4 Bc7
    17. Qg2 Be6 18. Nd2 O-O-O 19. Nxe4 dxe4 20. Bxe6+ Qxe6
    21. Rxe4 Qf7

Computer says it is 0, but my evaluation is that black has a secure king and already started Kingside attack, while white king is almost opened up and white has not attack.

  • 2
    This is only one single line taken into consideration, and I see no reason to believe it's forced. As is indicated in PhishMaster's answer, White doesn't need to play d2-d4 at once, it's perfectly fine to just go for d3 and eliminate the pawn on f4. In the best lines against the King's Gambit, white struggles but after 3...c5, white has quite an easy game. So I would say that the move is not particularly good.
    – Scounged
    Commented Feb 22, 2020 at 13:54
  • This continuation is awful
    – David
    Commented Mar 31, 2020 at 10:05

3...c5 is terrible here. It weakens the d6 and d5 squares and leaves black with a backward d-pawn. White will play d4 at some point leaving black with an IQP and white with a superior center. Black has no way to equalize against reasonable play. Typically, in a Sicilian black doesn't play e5 because holding back the e-pawn gives black potential control over d6 and d5. The one exception would be the Boleslavsky pawn formation however you'll notice that only works if black can keep the e pawn at e5 (because of the resulting endgame).

As white I would just play Bc4 followed by 0-0. If black plays Nf6 then e5 and there isn't anything black can do. After that, decide if you want to play the immediate d4 or play c3 followed by d4. Stockfish seems to think black is okay trying to hold the pawn with g5 but I see this as a position where black is going to have to fight very hard just to get a draw. In my opinion, this is what you hope for as white. You have an active position that gives black plenty of chances to go wrong but if black plays perfectly you should still be able to draw.


There is no obvious refutation other than white wins in the end.
Certainly not a single move !! as shown in the tournament book recap.

Chess Tempo shows black doing well with it but those games are from all players 2200 and down. Looking at some of them, white was nowhere near 2200 with the moves they made.

The key point is that nobody above 2200 plays c5 as black. [maybe a few do in a bigger database] So the refutation comes down more to the players ability to beat an inferior move.

With no top GMs showing the way you will need to use a good engine to find the refutation. But I suspect it will be too subtle for you to duplicate OTB when your opponent makes yet another deviation from the normal moves.

Following good chess principles and developing and then improving whites position until tactics decides or you win the end game would be a good strategy.

I never made GM so while I would be tempted to play b4 here I would likely just play nb3 bb4 and then pq3 to have a better position and edge in development while black has awkward pawns and a king in a weak position.

  • You had me confused with nb3 and bb4 and pq3 but I think I figured it out—is that descriptive notation? But then what is "b4"?
    – bof
    Commented Feb 22, 2020 at 14:19
  • mostly descriptive. b4 was algebraic. the other would be n-c3 b-c4 and d3 or p-d3. sorry but i think in descriptive and sometimes do not translate unless asked. Commented Feb 22, 2020 at 14:21
  • and the cowards are again busy downvoting a correct answer Commented Feb 22, 2020 at 18:27
  • 1
    I don't think this reply deserves a downvote necessarily, but some edits to tidy things up are in place. In particular I think that the mixing of algebraic and descriptive notation in the last paragraph should be edited to algebraic, since that is the notation used in other parts of the answer. When it comes to finding a refutation, I don't think there is a concrete "refutation" in the sense that black loses, but it's not very difficult to refute 3...c5 on a purely conceptual level; it basically wastes a move to prevent the d4 break, which is not something that white prioritizes anyway.
    – Scounged
    Commented Feb 23, 2020 at 0:41
  • possibly. but i am olde and i think in descriptive. clearly blacks c5 is bad and it should lose but in practice GMs do not play it and the weaker players who do win a lot of the games per chess tempo although they may lose more in other databases. Commented Feb 23, 2020 at 1:17

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.