How to Play the Italian Game: Blackburne-Kostić Gambit

The Blackburne-Kostić Gambit is all about activity and initiative. By playing 3...Nd4, you invite White to capture your knight with 4.Nxd4 (the engine's best move), which opens the centre after 4...exd4. In return for the piece, you get a pawn on d4, quick development after ...g6, and long-term pressure down the e-file. White's king may feel awkward if they castle short into your bishop's diagonal, or they may have to spend time defending the d4 pawn. The engine evaluates the position at +1.22, a clear edge for White — so the computer thinks White should eventually consolidate. But the practical statistics tell a different story: across over 8 million games, Black wins 49.9% of the time, while White wins 46.6%. That's a fantastic result for a gambit that the engine dislikes on paper. Your goal is to make White prove they can handle the chaos.

Stockfish's top recommendation is 4.Nxd4, which leads to 4...exd4 5.O-O g6. After 5.O-O (castling into safety and threatening the d4 pawn), Black's best continuation is 5...g6, preparing to fianchetto the bishop and put immediate pressure on the long diagonal. This is the critical test of your gambit: can White contain your activity while nursing their extra pawn? The engine believes yes — hence the +1.22 evaluation — but in practice, this is a tricky line for White to handle. You'll aim to develop your pieces quickly and keep the tension high. Remember: Black scores 49.9% overall from this position, so even in the main line you're far from lost.

Many White players don't know the correct response and grab the wrong pawn. Three inaccuracies stand out in the data: Nxe5 (1,866,158 games, White scores only 22.7%), c3 (922,988 games, White scores 52.8%), and d3 (716,045 games, White scores 54.1%). Nxe5 is the worst offender — it costs White about 0.9 pawns of advantage compared to the correct Nxd4. That 22.7% score for White is dreadful — practically a gift for Black. The moves c3 and d3 are also inaccurate (losing roughly 0.8 and 0.7 pawns of advantage respectively), as they try to kick or blockade the knight on d4 but fail to punish it. In both cases, Black can maintain the knight on d4 or retreat with tempo, keeping excellent play. The key takeaway: when White plays anything except 4.Nxd4, you're already on top.

The 8 million games in the database offer a fascinating picture. On the surface, White's best try is 4.Nxd4 (scoring 54.7%), followed by sidelines like O-O (55.6%) and d3 (54.1%). But look closer: 4.Nxd4 is also the most popular by far (over 3.2 million games), meaning most White players do choose the engine's move — and they still only score 54.7%. Meanwhile, after 4.Nxe5, White's score plummets to 22.7%, meaning Black wins more than 3 out of 4 games in that line! If you encounter 4.Nxe5, you should be smiling. The Blackburne-Kostić Gambit thrives on the fact that even strong moves for White lead to complex positions where Black has easy, intuitive play, while White's best continuation requires precision to convert.