Chess Opening Theory/1. e4/1...e5/2. f4/2...exf4/3. Nf3

This develops a knight, controlling the h4 square in order to prevent black from playing Qh4+, forcing white to play Ke2. White couldn’t block by g3??, because black can capture by fxg3! . Plus, white couldn’t capture back, since the h-pawn is pinned to the rook, forcing white to play Ke2.

There is a huge variety of possible approaches for Black here.
 * 3...g5 is the Classical Variation, and unlike a lot of other Classical Variations it's still the most common move. The purpose is twofold: protect the marauding f4-pawn in the event that White attacks it by moving the d-pawn, and threaten to advance to g4 and kick the knight away from f3. The weakening of Black's kingside is not too important a factor, because the f4-pawn gives Black all the space on that wing. Any kingside advance from White will run into Black's kingside advance coming the other way.


 * 3...d5 is the Abbazia Defence or Modern Defence. Black reasons that a less dramatic way of reinforcing the f4-pawn is by playing ...Bd6, but the immediate 3...Bd6 would block the d-pawn and get hit by an e5 advance, so Black gets that pawn out of the way first. Unfortunately, there's a White pawn on e4 ready to capture it, so it's not quite that simple, and the Abbazia is one of the few King's Gambit lines that score better for White than Black.

Those are the three mainstream options, then there are another four somewhat respectable options: Minor lines include 3...f5 and 3...h5.
 * 3...d6 is the move with which Bobby Fischer claimed to have refuted the entire King's Gambit. Black would still like to play ...g5 next move, but first takes the precaution of denying the f3-knight access to e5. This means it's going to be short of good squares after a g7-g5-g4 advance, and may have to make an abject retreat to g1. The Fischer Defence would probably be more popular still if bringing the knight to e5 in the 3...g5 variations actually scored well for White, but it doesn't.
 * 3...Be7 is the Cunningham Gambit. Black is determined to get a check in on h4! But White gets to make another move first, and after 4.Bc4 the check is much less powerful because the king just steps across to f1. The usual response to 4.Bc4 is 4...Nf6, but then what's the point of having the bishop on that diagonal?
 * 3...Ne7 is the Bonsch-Osmolovsky Variation, a new idea that's still rocketing up the popularity charts. The idea is to play ...d5 with an improved version of the Abbazia Defence, because if White exchanges pawns Black will get the knight to d5 from where it defends f4. This is scoring incredibly well even by King's Gambit standards: 49% Black wins to 22% White. A note of caution: this may simply be because the players on the White side of it haven't been very good, because in the 21st century the very good players tend not to play the King's Gambit.
 * 3...h6 is the Becker Defence, which prepares to simply defend the forward f-pawn with ...g5. How is this different from playing 3...g5 immediately? Because of the effects of White's undermining move h4. After 3...g5 4.h4 Black would be unwise to try to defend g5 with 4...h6? because after 5.hxg5, that h6-pawn is going to be pinned to the h8-rook. So Black needs to play two moves at once: 4...h6 to defend the g5 pawn, and 4...Bg7 to protect the rook. You're not allowed to play two moves at once, and you can't put a bishop on g7 while there's still a pawn there, so the only way of making it all work is to play ...h6 in advance. Simple!
 * 3...Nf6 is the Schallopp Defence. This is no straightforward developing move, because the knight is going to get kicked straight off f6 by White's e-pawn. The square Black has in mind for it is h5! That Black can spend two moves putting a knight on the edge of the board and still win 38% of games is probably a bad omen for King's Gambit players.

Theory table
.

1.e4 e5 2.f4 exf4 3. Nf3