Two Knights Defense

After 1.e4 e5 2.Nf3 Nc6 3.Bc4, the Italian Game's main fork in the road is move three for Black. The Giuoco Piano answers with 3...Bc5, developing the bishop to a square that eyes f2 and mirrors White's own setup. The Two Knights Defense does something more provocative: 3...Nf6 develops a second knight and attacks e4 directly, ignoring the bishop's diagonal entirely. Where the Giuoco Piano is a discussion about who controls the center more comfortably, the Two Knights is an argument about tempo — Black is daring White to prove that 4.Ng5, grabbing at f7, is actually good before White's own king is anywhere near safe.

Why 4.Ng5 is the whole point of the opening

Nothing about the Two Knights makes sense without 4.Ng5. The move looks crude — a developed piece retreating half its progress to attack a pawn — but it works only because Black's last move left f7 defended solely by the king, and White is betting that grabbing it, or the complications around it, outweighs the awkwardness of moving one piece twice in the opening. Black's most reliable reply, 4...d5, meets the threat by opening the center immediately: after 5.exd5, Black no longer has to worry about a quiet continuation and must choose between several committal recaptures, each one branching into its own middlegame rather than converging on a shared main line.

The fork between two entirely different fourth moves

5.exd5 Na5 is the modern main line: the knight steps to the rim, attacking the c4 bishop and accepting a temporarily awkward piece for the promise of winning it back with an improved structure once White's knight on g5 has to retreat or commit to a piece sacrifice. The sharper 5...Nxd5 leads directly into the most theory-dense corner of the whole opening, where White can try 6.Nxf7, sacrificing a piece to drag the king into the center — that specific idea has its own dedicated treatment elsewhere — it has almost nothing in common with the rest of the Two Knights and deserves to be learned as its own unit rather than folded into a general overview.

The other two replies to 4.Ng5, and why they aren't the main line

5...Na5 and 5...Nxd5 aren't the only tries. 5...Nd4, the Fritz Variation, offers a pawn to establish the knight on a strong central square and meets 6.c3 with 6...b5, and 5...b5, the Ulvestad Variation, offers the pawn outright to gain time on the bishop before White can consolidate. Both are playable and both carry real, specific theory of their own, but neither has displaced 5...Na5 as the main practical choice, mostly because Na5's structural logic — win back the piece, keep a sound pawn skeleton — is easier to execute reliably at the board than the more speculative gambits are. A club player who meets 4.Ng5 for the first time with one of these instead of the two main tries isn't doing anything wrong, but should know that far less has been written about defending them precisely, which cuts both ways: less preparation is needed to play them, and less preparation is available if the opponent knows a strong answer.

What White gives up by avoiding 4.Ng5

Plenty of White players sidestep the whole argument with 4.d3 or 4.O-O, developing normally and accepting a slower, more strategic game instead of a forced tactical one. This isn't a lesser try — it simply changes what the opening is about. Without the immediate f7 threat, Black's 3...Nf6 loses most of its bite as a provocation and the game more closely resembles a standard Italian, with the practical difference that Black's knight is already committed to f6 rather than kept flexible. Players who choose the Two Knights specifically for its sharpness are often disappointed by an opponent who declines the fight, and that mismatch of expectations is itself a common source of aimless play from Black once move four turns out to be quiet.

Black often treats 4...d5 as automatic without having actually decided in advance between 5...Na5 and 5...Nxd5. These lead to different structures and different amounts of danger, and improvising the choice at the board, rather than having studied at least one of them properly, is what most often wrecks the second player's game in this opening.

For the calmer, mirroring alternative on move three, see the Italian Game overview. For White's other pawn-sacrifice idea against the same 3...Nf6 setup, compare the Evans Gambit, which offers material against 3...Bc5 instead of chasing f7.