Some chess moves are brilliant. A rare few transcend the game itself. Alexei Shirov's 47...Bh3!! against Veselin Topalov at Linares 1998 is one of them — a bishop sacrifice so counter-intuitive that chess engines of the era evaluated it as a losing blunder, and even Garry Kasparov, watching from the ringside, couldn't immediately explain what Shirov had seen.
This game is routinely cited by grandmasters as containing "the greatest endgame move ever played." The move appears to give away a bishop for nothing — it's attacked by two pawns and defended by none — yet it creates a decisive zugzwang that forced Topalov to resign 20 moves later. To understand why 47...Bh3!! is so special, we need to walk through the full game, the players, and the deep positional logic behind the sacrifice.
Play through the complete game below. The critical moment comes at move 47:
Topalov vs Shirov, Linares 1998 — Play through the full game from the Grünfeld opening through the endgame. The critical position arrives at move 47.
The Players and the Tournament
The 1998 Linares super-tournament was one of the strongest events of its era. If you want to train your pattern recognition to spot creative moves like Shirov's, studying super-tournament games is one of the best approaches. Veselin Topalov (White, 2740 Elo) was already an established world-class player, known for his aggressive, attacking style and deep preparation. He would go on to become FIDE World Champion in 2005. Alexei Shirov (Black, 2710 Elo) was one of the most creative and dangerous players on the circuit — a Latvian-born grandmaster famous for his tactical imagination and willingness to sacrifice material for positional compensation.
Linares was the Wimbledon of chess — an invitation-only round-robin featuring the world's elite. The tenth-round pairing between Topalov and Shirov came at a crucial point in the tournament. Both players needed a win, and neither was inclined to offer a quick draw. The game began as a Grünfeld Defense, Exchange Variation — a sharp opening choice that signaled Shirov's fighting intentions from move one.
Opening: The Grünfeld Exchange (Moves 1–14)
The Grünfeld Defense (1.d4 Nf6 2.c4 g6 3.Nc3 d5) is named after Austrian GM Ernst Grünfeld. Understanding opening principles helps you appreciate why Black's setup is so strategically ambitious. Its core idea is radical: Black invites White to build a strong pawn center, then attacks it with pieces, hoping to prove that central pawns are targets rather than assets. The Exchange Variation (4.cxd5 Nxd5 5.e4 Nxc3 6.bxc3) is the most direct approach — White accepts the central space and challenges Black to prove compensation.
1.d4 Nf6 2.c4 g6 3.Nc3 d5 4.cxd5 Nxd5 5.e4 Nxc3 6.bxc3 Bg7
At this point, White has a dominant pawn duo on d4 and e4, while Black has a fianchettoed bishop on g7 bearing down on the center. This is the classic Grünfeld trade-off: White's space versus Black's activity.
7.Bb5+ c6 8.Ba4 O-O 9.Ne2 Nd7 10.O-O e5 11.f3
White's setup is solid but slightly passive. The bishop on a4 is outside the pawn chain, and White has committed to f3, reinforcing the e4 pawn but temporarily blocking development of the g1-knight's natural f3 square. Black's knight on d7 is headed for c5 or b6 to pressure the center.
11...Qe7 12.Be3 Rd8 13.Qc2 Nb6 14.Bb3 Be6
Both sides complete development. White's pieces are well-placed for central control, while Black's pieces target the d4 and c4 squares. The position is balanced, with White's space advantage offset by Black's active piece play.
Position after 14 moves — White has a comfortable space advantage but Black's pieces are actively placed. The battle for the center continues.
Middlegame: Maneuvering and Counterplay (Moves 15–30)
15.Rad1 Nc4 16.Bc1 b5 17.f4 exd4 18.Nxd4 Bg4 19.Rde1 Qc5+ 20.Kh1
White's play centers on the f4-f5 pawn break to open lines against Black's king. This kind of middlegame planning — identifying the critical pawn break — is what separates strong players from passive ones. Black counters by pressuring the c3 pawn and the d4 knight. The position is sharp but balanced — one wrong move could tip the scales.
20...a5 21.h3 Bd7 22.a4 bxa4 23.Ba2 Be8
Black's 22...bxa4 is a critical decision. By giving up the b-pawn, Shirov opens the b-file for his rook but also creates a passed a-pawn for White. This is typical Shirov — accepting structural weaknesses in exchange for dynamic counterplay.
24.e5 Nb6 25.f5 Nd5 26.Bd2 Nb4 27.Qxa4 Nxa2 28.Qxa2 Bxe5
Shirov sacrifices the exchange of a knight for a bishop — not a conventional trade, but one that leaves White's light-squared bishop gone. In opposite-colored bishop endgames, this is a subtle advantage. Already, Shirov is steering the game toward an endgame where his bishop will be the more active piece.
29.fxg6 hxg6 30.Bg5 Rd5
White's attack has dissipated. The queenside passed pawn has been liquidated, and Black's pieces are active. The game is transitioning into an endgame where Black's bishop pair and active rooks give him full compensation for the pawn deficit. Shirov has outplayed Topalov in the middlegame's second act.
Endgame Transition (Moves 31–40)
31.Re3 Qd6 32.Qe2 Bd7 33.c4 Bxd4 34.cxd5 Bxe3 35.Qxe3 Re8
Both sides exchange into a queen-and-pawn endgame. Endgame patterns like these occur frequently in club games — recognizing the transition point is critical. White has an extra pawn but Black's pieces are more active. The isolated pawn on d5 is weak, and Black's rook on e8 controls the open file. Shirov's positional pressure is beginning to tell.
36.Qc3 Qxd5 37.Bh6 Re5 38.Rf3 Qc5 39.Qa1 Bf5 40.Re3 f6
Topalov's position is slipping. The d5 pawn fell, and now Black's pieces are coordinating beautifully. The bishop on f5 controls key diagonals, the rook on e5 is aggressively placed, and the queen on c5 supports both offense and defense. White's pieces, by contrast, are passive.
Position after 40 moves — Black's pieces dominate. The rook on e5, queen on c5, and bishop on f5 outgun White's passive setup. The endgame is slipping away from Topalov.
The Final Maneuvers (Moves 41–47)
41.Rxe5 Qxe5 42.Qa2+ Qd5 43.Qxd5+ cxd5 44.Bd2 a4 45.Bc3 Kf7 46.h4 Ke6
Rooks come off, and the game enters a pure opposite-colored bishops endgame. King-and-pawn endgame technique is essential here — knowing when to centralize your king can make or break the result. Black has a passed d-pawn and a4-pawn, while White has a pawn on h4 and a kingside majority. Opposite-colored bishops endgames are notoriously drawish — the defending bishop can blockade passed pawns on its color — but here, Black's king activity gives him a decisive advantage. Shirov's king is already heading to the center, while Topalov's king is still on g1.
47.Kg1
This is the critical mistake. Topalov plays a natural-looking king move, centralizing toward the queenside. But it's one tempo too slow. As we'll see, White's king needed to be on f2 immediately to hold the position.
Position after 47.Kg1 — White's king centralizes one tempo too slowly. Black to play and win with the most famous endgame move in chess history.
47...Bh3!! — The Move That Broke Chess
"The best endgame move ever played." — Various grandmasters
Learn more about what makes a brilliant move in chess and how FireChess classifies them.
Shirov played 47...Bh3!!, placing his bishop on a square attacked by White's g2-pawn and adjacent to the h4-pawn. On the surface, it looks like a simple blunder: both pawns can capture the bishop for free. Early computer engines evaluated it as simply losing — a bishop for nothing is a terrible trade by any material calculator.
But the sacrifice is the point. The move is a profound example of tempo counting in king-and-pawn endgames. Let's examine both responses:
47...Bh3!! — The bishop is en prise to two pawns. After 48.gxh3, Black's king marches to e4 one critical tempo ahead of White's, and the d-pawn becomes unstoppable. Play through the main line and watch the badges.
If White Captures: 48.gxh3
This is the critical test. If Topalov captures the bishop immediately, Black's king exploits the cleared path:
48.gxh3 Kf5 49.Kf2 Ke4
The black king reaches e4 while White's king is still on f2. That single tempo — gained by forcing White to capture with the g-pawn instead of moving the king to f2 — is the entire difference between winning and drawing. From e4, Black's king blocks the d5-pawn's path forward and supports its advance. White's bishop on c3 can't stop the pawn because... well, let's see.
50.Bxf6 d4 51.Be7 Kd3 52.Bc5 Kc4 53.Be7 Kb3
The black king marches to b3 to support the a4-pawn. White's bishop is stuck trying to stop both the d-pawn and the a-pawn simultaneously. The white king is still tied down on the kingside, guarding the h3-pawn. Black's two passed pawns, supported by the king, are unstoppable. White must resign.
The key insight: White's g-pawn was in the way of his own king's path to the center. By forcing White to capture with the g-pawn, Shirov's sacrifice clears the diagonal and costs White one critical tempo. In king and pawn endgames, one tempo is the difference between winning, drawing, and losing. This is that tempo.
If White Declines: 48.Kf2
Topalov could also try to simply ignore the sacrifice and bring his king toward the center:
If White declines with 48.Kf2, Black takes the g2-pawn and creates three connected passed pawns on the kingside — an even faster win.
48.Kf2 Kf5 49.Ke3 Bxg2
Now Black simply captures the g2-pawn with the bishop, giving Black three connected passed pawns (f6, g6, h3) on the kingside. White's bishop can't halt all three, and Black's king has complete freedom. This line is actually even worse for White — Black converts with minimal resistance, and the game would end even faster than the capture line.
Why Computers Couldn't See It
In 1998, chess engines evaluated positions almost entirely based on material count. If you've ever wondered how to read chess engine analysis, this game is a masterclass in why raw evaluation numbers can mislead. A bishop is worth roughly three pawns. Sacrificing a bishop with no immediate compensation — no check, no capture, no direct threat except the vague positional idea of "king activity" — registered as a catastrophic blunder in every engine.
The reason the sacrifice works is purely positional: it exploits the unique geometry of opposite-colored bishops in king-and-pawn endgames. In such endgames, the bishops operate on different color complexes, so an extra pawn often means nothing if the enemy king can blockade. But here, the key factor is king activity — specifically, whose king reaches the center first.
Modern engines like Stockfish 18 evaluate 47...Bh3!! at roughly -5.0 (decisive for Black) after just a few seconds of search. They can see the full 10-ply king march that humans calculate by intuition. The material loss is irrelevant because the tempo gain is decisive.
This is one of the earliest and most famous examples of a positional sacrifice that computers completely misevaluated. It helped drive the development of more nuanced evaluation functions in chess engines — functions that understand king activity, tempo, and pawn structure beyond simple material counting.
The Aftermath: Shirov's King March
Topalov chose 48.gxh3, the capture line. This kind of concrete calculation training — working through forced lines move by move — is how you develop the vision to find moves like 47...Bh3 in your own games. Over the next 6 moves, Shirov executed one of the most famous king marches in chess history:
| Move | White | Black | Significance |
|---|---|---|---|
| 48 | gxh3 | Kf5 | King steps into the cleared path |
| 49 | Kf2 | Ke4! | Critical — one tempo ahead of White's Kf3 |
| 50 | Bxf6 | d4 | Pawn advances, bishop can't stop both passed pawns |
| 51 | Be7 | Kd3 | King invades, supporting d-pawn |
| 52 | Bc5 | Kc4 | King shifts to support the a-pawn |
| 53 | Be7 | Kb3 | Both passed pawns unstoppable — White resigns |
The final position is beautiful in its clarity: Black's king on b3 supports the a4-pawn toward promotion, while the d4-pawn is also threatening to advance. White's bishop can stop one but not both. The white king on f2 is helpless, stuck guarding the kingside pawns.
Final position after 53...Kb3 — Black's king supports the passed a-pawn while the d-pawn also threatens. White's bishop can't stop both, and the king is tied down on the kingside. Topalov resigned.
Lessons for Club Players (1200–1800)
This game teaches several critical endgame concepts that directly apply to your own games. Scan your games on FireChess to see how your endgame accuracy compares — you might find the same tempo-counting mistakes Topalov made.
Q: King Activity Trumps Material in Endgames
This is the single most important lesson from Shirov's sacrifice. He gave up a full bishop — normally a decisive material advantage — for a single tempo. In the middlegame, this would be catastrophic. But in the endgame, especially in opposite-colored bishop endgames, the active king is worth more than a minor piece.
Practical application: When you reach an endgame where material is roughly balanced, actively ask yourself: "Which king is more active?" If your king is stuck defending pawns while the opponent's king is free to roam, consider sacrificing a pawn or even a piece to activate your king. The Shirov game proves that this trade can be winning.
Q: Opposite-Colored Bishops Favor the Attacker
When both sides have only one bishop each, and they run on different colored squares, the defender faces a structural problem: the defending bishop can never attack the opponent's pawns if they're on the opposite color. This means:
- Passed pawns on the opposite color from the defender's bishop are extremely dangerous
- The attacking player can often sacrifice material because the defender's bishop is irrelevant on key squares
- Drawish endgames can become winning with the right pawn structure
In the Shirov game, White's bishop was on c3 (a dark square), while Black's passed d-pawn and a-pawn were also on dark squares. White's bishop could theoretically attack them, but Black's king support made any capture impossible. Black's bishop, by contrast, had already done its job — after initiating the sacrifice, it was irrelevant.
Practical application: In opposite-colored bishop endgames, prioritize getting your passed pawns onto the opposite color from the enemy bishop. Create two passed pawns as far apart as possible — the defending bishop can only stop one at a time.
Q: Don't Trust Static Evaluations in King-and-Pawn Races
This game is a perfect example of why you should calculate concretely rather than trust material evaluation. Even a modern 2000-rated player glancing at a quick Stockfish evaluation might think "bishop for nothing? That's losing." But the concrete calculation proves otherwise.
The rule of thumb: In king-and-pawn endgames, count tempi, not material. Each move matters. If your king can reach a critical square one move before the opponent's king, that's often worth a pawn or more.
Practical application: When you see a move that gives away material but activates your king, don't dismiss it out of hand. Calculate the king-and-pawn race concretely. Count: "If I sacrifice here, how many moves does it take my king to reach the key square? How many moves for the opponent?" If your king arrives first, the sacrifice works regardless of the material count.
Q: Recognize When King Centralization Becomes Critical
In the opening and middlegame, king safety is paramount — you keep your king hidden behind pawns and pieces. But in the endgame, the king transforms from a liability into a fighting piece. The transition point usually comes when queens are off the board and fewer than 5 pieces remain per side.
Practical application: Develop a mental checklist for endgame transitions:
- Are queens off the board?
- Are there fewer than 10 pieces total?
- Is my king safe to leave its shelter?
- Can I reach the center before the opponent's king?
If you answer Y-Y-Y-Y, start marching your king forward immediately.
Q: Opposite-Colored Bishops Weaken the Defender — Exploit It
The Topalov-Shirov game is the ultimate demonstration of this principle. White's bishop on c3 was on dark squares, while Black's threats were on dark squares too — but the d4-pawn was on a dark square, meaning White's bishop could theoretically attack it. Except Black's king on c4 defended the pawn, and White's king was too far away to help.
The reason opposite-colored bishop endgames favor the attacker is simple: the defender's bishop cannot contest the attacker's bishop. If both bishops were on the same color, the defender could trade or block. But on opposite colors, each bishop rules its own domain.
How This Compares to Other Famous Endgame Sacrifices
| Game | Sacrifice | Idea |
|---|---|---|
| Shirov vs Topalov, 1998 | Bishop for tempo | King activity in opposite-colored bishop ending |
| Carlsen vs Karjakin, 2016 (WCC) | Exchange sacrifice | Active rook vs passive bishop |
| Karpov vs Kasparov, 1985 (WCC) | Pawn sacrifice | Space advantage and piece activity |
| Fischer vs Taimanov, 1971 | Exchange sacrifice | Bishop dominates knight in closed position |
What makes Shirov's sacrifice unique is the purity of the idea. For more famous tactical moments, explore chess gambits for club players or study chess opening traps that punish material greed.: a bishop for one tempo, with no hidden material compensation. In the other examples above, the sacrificer gets something tangible (space, piece activity, structural damage). Shirov got nothing except... the right to move his king first. And that was enough.
See the Complete Game
You can play through the complete Shirov vs Topalov 1998 game on FireChess with our interactive board. It's one of the most-visited famous games on our site — and now you know why that bishop on h3 is so special.
More Chess Improvement Resources
- Analyze Your Own Games Free — FireChess scans your Lichess or Chess.com games with Stockfish 18 to find tactical patterns you miss
- Chess Endgame Patterns Club Players Miss — The zugzwang and king-activity patterns that win games
- Average Centipawn Loss by Rating — Benchmark your accuracy against players at your level
- Guess the Elo — Test your positional judgment against real games, learn how to estimate rating from any PGN, or read our guide to estimating rating from positions
Other Famous Sacrifices on FireChess
- The Game of the Century — Byrne vs Fischer, 1956
- Kasparov's Immortal — Kasparov vs Topalov, 1999
- The Immortal Zugzwang Game — Sämisch vs Nimzowitsch, 1923
Frequently Asked Questions
Before diving in, make sure to scan your own endgames on FireChess to see how your accuracy compares to Shirov's precision.
Q: What was the final score of Shirov vs Topalov at Linares 1998?
The game ended with Shirov winning after 53 moves. Topalov resigned in the position after 53...Kb3, where Black's king supported the a4-pawn and the d4-pawn was unstoppable. The full game lasted 53 moves from a Grünfeld Defense, Exchange Variation. Shirov's victory was a crucial result at the Linares super-tournament, which featured the world's strongest players.
Q: Why is 47...Bh3!! considered the greatest endgame move ever played?
The sacrifice is extraordinary because it gives up a full bishop for zero immediate material compensation — the piece is attacked by two pawns and defended by none. What makes it immortal is the depth of the idea: by forcing White's g-pawn to capture on h3, Shirov gains exactly one tempo in the king march to e4. That single tempo is the difference between a draw and a win. No other famous endgame sacrifice is as pure — it trades a bishop solely for the right to move first in a king-and-pawn race.
Q: Why couldn't chess engines evaluate 47...Bh3!! correctly in 1998?
In 1998, engines like Fritz and Rebel relied heavily on material counting. A bishop is worth roughly 3 pawns, so sacrificing it with no check, capture, or direct threat registered as a catastrophic blunder. The sacrifice works for purely positional reasons — king activity and tempo in an opposite-colored bishop endgame — concepts that early engines lacked the evaluation depth to recognize. Modern engines like Stockfish 18 instantly evaluate it as winning for Black at roughly -5.0.
Q: What are the key themes to learn from the Shirov-Topalov endgame?
Four major endgame principles emerge: (1) King activity trumps material — in endgames, an active king can be worth more than a minor piece; (2) Count tempi, not material — in king-and-pawn races, the side whose king arrives first wins regardless of material; (3) Opposite-colored bishops favor the attacker — the defending bishop cannot contest threats on its opposite color; (4) Static evaluations lie — always calculate concretely in king-and-pawn endgames rather than trusting material count.
Q: Has a bishop sacrifice like 47...Bh3!! been played in other grandmaster games?
While bishop sacrifices for positional compensation occur occasionally at the GM level, Shirov's 47...Bh3!! remains unique in its purity. Other notable endgame sacrifices include Carlsen's exchange sacrifice vs Karjakin (2016 World Championship) and Fischer's exchange sacrifice vs Taimanov (1971 Candidates). However, none match the elegance of trading a full bishop for a single tempo with no residual material gain. The sacrifice has since become a standard teaching tool in endgame training courses worldwide.
What makes a move immortal? It's not just the outcome — it's the idea behind it. Shirov's 47...Bh3!! was a move that defied material logic, computational evaluation, and even grandmaster intuition. It's a reminder that chess, at its highest level, remains an art form — one where a single bishop sacrifice can earn its place in history forever.
Since this game, no serious endgame training session is complete without studying the Bh3 sacrifice. It has become a standard teaching tool for understanding king activity, tempo counting, and the unique dynamics of opposite-colored bishop endgames. Every club player should work through this position at least once.
If you want to see how your own endgame decision-making compares to Shirov's, scan your games with FireChess and check your endgame accuracy. You might be surprised how often the right move is the one that looks wrong on material alone.