The Immortal Game: Anderssen v Kieseritzky, London 1851

In 1851, during the London tournament, Adolf Anderssen and Lionel Kieseritzky played an informal game in a break between rounds. Anderssen gave up a bishop, both rooks and finally his queen, and mated with a bishop and two knights. Kieseritzky reportedly sent the moves to his chess club in Paris, and the magazine La Régence published them that July. In 1855 Ernst Falkbeer gave the game the name it has kept ever since: the Immortal Game.

The final position, after 23.Be7#. Black has a queen, two rooks and a bishop more, and it's mate.

The whole game

The Immortal Game: Anderssen v Kieseritzky, London 1851. Step through all 23 moves.

The opening is the King's Gambit, 2.f4, one of the most popular openings of the day. After 3...Qh4+ White's king has to move and can no longer castle. 4...b5, the Bryan Countergambit, is a pawn sacrifice few players today consider sound. Count the queen moves: eight of Black's first twenty moves are made by the queen, while White brings out the other pieces.

What Stockfish makes of it

The Immortal Game is famous for its sacrifices, but they weren't all sound, and for a long stretch it was Black who stood better. We had Stockfish check every move.

11...cxb5: the turning point

Black to move after 11.Rg1. With 11...h5, Black is better.

By 10.g4 White was worse: Stockfish rates Black about a pawn better. Here Kieseritzky took the bishop with 11...cxb5. Stockfish prefers 11...h5, which keeps Black better by more than half a pawn; after the capture, White is better by a pawn and a half. Robert Hübner, who analysed the game closely, also called 11...cxb5 Black's critical mistake and recommended 11...h5.

12.h4, 13.h5, 14.Qf3: only one move each time

11...cxb5, then three moves where only one keeps White's attack going: 12.h4, 13.h5 and 14.Qf3.

What follows is less famous than the sacrifices and more precise. At each of White's next three moves, only one move keeps the advantage: 12.h4, 13.h5 and 14.Qf3. Every other move gives it away: at moves 12 and 13 the next best leaves White a pawn worse, and at move 14 it only holds the balance. This is the kind of moment Master moves trains: one move works, and you have to find it.

18.Bd6: the famous sacrifice

18.Bd6: Anderssen lets both rooks go.

18.Bd6 offers the rook on g1, and 19.e5 then leaves the rook on a1 to the queen as well. It's the move that made the game, and it isn't the strongest. Stockfish prefers 18.Re1, about four pawns better for White, with 18.d4 close behind. After 18.Bd6 White is still better, but only by about a pawn and a half. Hübner gave three moves he thought were better: 18.d4, 18.Be3 and 18.Re1. Garry Kasparov has remarked that if the game had gone the unspectacular way, chess would have lost a crown jewel.

Black didn't take the best way out either. Wilhelm Steinitz suggested in 1879 that 18...Qxa1+ was better than 18...Bxg1, and Stockfish agrees: White stays better by about a pawn and a half after 18...Qxa1+, against about four pawns after 18...Bxg1. Even 19.e5 isn't Stockfish's first choice (19.Re1 was stronger), but it keeps White well on top.

20...Na6: one move holds, and it isn't this one

Black to move after 20.Ke2. Only 20...Ba6 avoids mate.

Black is two rooks and a bishop up, and White's pieces are all aimed at Black's king. Stockfish finds exactly one defence: 20...Ba6, which clears c8 so the king can escape there. White stays better, by about two pawns, but has work to do. Every other move, 20...Na6 included, allows a forced mate.

20...Na6 21.Nxg7+ Kd8 22.Qf6+ Nxf6 23.Be7#.

Did it really end like this?

Perhaps not on the board. Kieseritzky reported that he resigned after 20.Ke2. An article by Friedrich Amelung from 1893, cited by Hübner, says he more likely played 20...Na6, after which Anderssen announced the mate. Either way, the mating finish is how the game has been published ever since.

How we checked

The moves are from Wikipedia's page on the game. We replayed every one with python-chess: all are legal and the game ends in checkmate. Older books print the game with different move orders, and some versions contain errors.
The engine: Stockfish 16.1 checked every move at depth 22, and the key moments again at depth 28. "A pawn better" is Stockfish's measure of the position, not a count of material.
The history and the named opinions (Hübner, Steinitz, Kasparov, Amelung) are from the Wikipedia article.

Try it. Real positions, set at your level, every answer checked against the board. No account needed to start.

Start training