What Is a Sacrifice in Chess?
Spielmann's real vs sham distinction, the six types (positional, attacking, defensive, clearance, deflection, exchange), four ways sacrifices pay you back, seven famous examples from Anderssen to Carlsen, and the five mistakes that turn every legal capture into a losing gamble. No signup. No paywall.
TL;DR — Sacrifice in 5 Bullets
- A sacrifice is a deliberate voluntary offer of material for a bigger dividend — tempo, initiative, king exposure, or a permanent positional asset.
- Spielmann's 1935 typology: real sacrifices never return the material (positional compensation); sham sacrifices recover everything within a forced sequence (tactical).
- Six types: positional, attacking, defensive, clearance, deflection, exchange. Each has a distinct dividend and a distinct calculation method.
- Every sacrifice needs a concrete follow-up. If you can't name 3-5 forced moves after the sacrifice, it's a blunder, not a sacrifice.
- Modern engines endorse positional exchange sacrifices more than any previous generation of theory — the Sicilian and King's Indian Rxc3 lines are all engine-approved.
What Does “Sacrifice” Mean in Chess?
A chess sacrifice is the deliberate voluntary offering of material — a pawn, minor piece, rook, or queen — for a specific bigger dividend. The word carries the same connotation as its everyday English use: you give up something valuable now for something more valuable later. The distinguishing feature is intent: an accidental piece loss is a blunder; the same move offered voluntarily with a concrete plan for compensation is a sacrifice.
The dividend can arrive immediately (mate in 4-8 moves) or a very long time later (a structural asset that decides the game 30 moves after the sacrifice). This is what separates a gambit (an opening pawn sacrifice for development and initiative) from a queen sacrifice for mate in three: both are sacrifices, but their dividends and their calculation requirements are totally different.
The definitive typology comes from Austrian grandmaster Rudolf Spielmann's 1935 book The Art of Sacrifice in Chess. Every serious sacrifice-instruction book since — Vukovic's Art of Attack, Nunn's tactical primer, Aagaard's Attacking Manual — builds on Spielmann's framework. Start there.
Most attacking sacrifices are the front-piece move of a battery — the Greek Gift Bxh7+ is the bishop-front of a queen-plus-bishop battery aimed at the h-file; the Nezhmetdinov Rxb7 sac is the rook-front of a queen-plus-rook battery. Read the battery hub to see the geometric setup that makes every one of these sacrifices sound.
The Six Types of Sacrifice
Every sacrifice belongs to one of six categories, each defined by its dividend and its calculation method:
Positional sacrifice — trade material for a permanent structural asset
The material never comes back and the position never becomes tactical — instead the sacrificing side accepts the deficit forever in exchange for something structural: a passed pawn that cannot be stopped, a knight outpost the defender can never dislodge, a blockaded weak square, or the bishop pair in an open position. Petrosian's 1959 exchange sacrifice against Reshevsky (…Rxc3 for a permanent dark-square grip) is the textbook example — Petrosian never recovered the exchange, but the c3 knight and the a1–h8 diagonal decided the game 30 moves later. Positional sacrifices are the hardest to play and the hardest to punish because there is no tactical checkpoint that says 'the sacrifice worked'.
Attacking sacrifice — burn material to reach the enemy king
The classic Romantic-era sacrifice: give up a bishop, knight, or rook to strip the pawn shield off the enemy king, then deliver checkmate or a decisive material win within a fixed number of moves. The Greek Gift (Bxh7+ Kxh7 Ng5+ Kg8 Qh5) sacrifices a bishop to force the king to h7 and open the h-file with tempo. Tal's endless piece sacrifices in the 1960 World Championship match against Botvinnik — Nb5 in game 6, exchange sac in game 17 — are all attacking sacrifices where the return arrives as a mating net inside 8-12 moves. Every attacking sacrifice must have a concrete calculated end: without the calculation, it's just a blunder.
Defensive sacrifice — give material to survive an attack
Return material — often exactly the amount you're up — to blunt an incoming attack, trade queens off the board, or force a drawn endgame. Karpov's 1974 exchange-back-sacrifice against Korchnoi (returning a whole exchange to reach a technically drawn rook ending) and Petrosian's positional exchange returns against Botvinnik in 1963 are the two most-cited examples. In the modern engine era, giving back material to reach an opposite-coloured-bishop endgame is a standard defensive-sacrifice tool at every rating level.
Clearance sacrifice — remove your own piece from a critical square
Sacrifice a piece specifically to vacate the square it stands on, so a different piece can occupy it with tempo. Réti's mating pattern Kg2 Qg3+ Kh1 Qxg2#? No — 1.Rxh7+! (clearance sac of the rook so the queen can deliver mate on the newly opened diagonal). Every study composer since Troitsky has used clearance sacrifices as a building block. In practical play, the classic pattern is a rook sac on e5 or d5 to clear the file for the queen behind it.
Deflection / decoy sacrifice — force an enemy piece off a critical square
Offer material so the enemy piece MUST capture and, in capturing, leaves a square undefended. The most common tactical sacrifice type in the middlegame — every removal-of-the-defender combination is a deflection sacrifice. Anderssen vs Kieseritzky 1851 (the Immortal Game) is a chain of deflection sacrifices: 22.Qxg7+! Rxg7 removes the rook's defence of a mating square. Deep Blue's 1997 Nb5 sacrifice against Kasparov (game 4 turned game 6) was a deflection sacrifice — Kasparov's king had to move off the defence.
Exchange sacrifice — trade a rook for a minor piece
The most common non-attacking sacrifice at master level: exchange (rook = 5, minor piece = 3, so you're 'giving' 2 points) for one of four dividends — a permanent dark-square grip like Petrosian, a passed pawn like Kasparov's Rxc3 against Karpov 1985 game 16, control of an open diagonal, or the elimination of an enemy defender. Modern engines endorse positional exchange sacrifices more often than any previous generation of theory: the classic Sicilian Najdorf …Rxc3 and the King's Indian …Rxc3 lines are both engine-approved at master level. Distinct from a queen or piece sacrifice because the material differential is small enough to be recouped strategically.
What a Sacrifice Buys You — The Four Dividends
Every sacrifice trades material for one of four dividends. Name the dividend before you sacrifice, or the sacrifice is a blunder:
Tempo and initiative
Every gambit and every attacking sacrifice trades material for tempo. The Smith-Morra Gambit gives up one pawn for two developing tempi and open lines; the Danish Gambit gives up two pawns for three attacking tempi. In the middlegame, the Greek Gift bishop sacrifice trades a piece for four attacking tempi (Bxh7+, Ng5+, Qh5, mate) — the defender has no time to bring the queen back to defend. When engine says +1.5 for the sacrificing side, it's usually reporting that the initiative is worth 1.5 pawns of eventual material recovery.
Structural damage to the enemy pawn shield
A piece sacrifice on h7, f7, or g7 strips the shield in front of the enemy king. The king now has an open file OR two open diagonals aimed at it OR both. Tal's queenside sacrifices in the 1960 match against Botvinnik traded pieces for structural king-shield damage that never healed — even after Botvinnik consolidated the material, the king was permanently open, and the game ended on a mating attack ten moves later. The key metric: after the sacrifice, how many diagonals and files point at the enemy king? Two or more = full compensation.
King exposure and mating net
Force the enemy king to a specific square where a calculated mating net closes. Every Legal's Mate, every smothered mate, every back-rank finish is a sacrifice that forces the king to the target square. The classic sequence: (1) sacrifice a piece for a check, (2) the king moves to the calculated square, (3) a second sacrifice or a quiet mating move closes the net. Verify the mating net with concrete calculation before the first sacrifice — the sacrifice is only sound if the mate is FORCED, not merely LIKELY.
Long-term positional dominance
The dividend arrives 20 or 30 moves later as a converted structural advantage — a passed pawn, a light-square weakness, a blockaded outpost. Petrosian's positional exchange sacrifices and Karpov's strategic pawn sacrifices show this pattern: no immediate tactic, just a slow constriction that ends in resignation. The rule from Spielmann's Art of Sacrifice (1935): positional sacrifices are correct only if you can name the specific structural feature that justifies the material investment.
Real vs Sham Sacrifice — Spielmann's Distinction
Rudolf Spielmann introduced the fundamental split in 1935. Learn this distinction once and every sacrifice you see (or play) becomes classifiable:
| Property | Real sacrifice | Sham sacrifice |
|---|---|---|
| Return of material | Never — you accept the deficit permanently | Fully — usually within 2-8 moves |
| Calculation depth required | 0-3 moves (structural judgement decides) | Deep and forced (all lines to mate or recovery) |
| Typical dividend | Structural — outpost, passed pawn, king exposure | Tactical — mate, decisive material win |
| Risk profile | Unfalsifiable — you feel the compensation for 20+ moves | Binary — either the calculation works or you're just down material |
| Textbook example | Petrosian vs Reshevsky 1959 exchange sac | Anderssen's Immortal Game queen sac |
Seven Famous Sacrifices Every Player Should Know
The sacrifice canon — study one game per week and by the end of a year you will have internalised the seven core patterns:
Anderssen's Immortal Game — sacrifice both rooks, queen, and win
Anderssen vs Kieseritzky, London 1851. Anderssen sacrifices both rooks, then the queen, then delivers mate with three minor pieces. 22.Qxg7+! Rxg7 23.Bxf6+ Ke8 (24.Bxg7 wins) — but Anderssen instead played 24.Bxg7# in the final move. Six sacrificial moves in a row. The founding romantic-attack sacrifice game — every 'sac everything' tradition traces back here.
Anderssen's Evergreen Game — queen sac Qxd7+
Anderssen vs Dufresne, Berlin 1852. 19.Rad1! prepares the queen sacrifice; 20.Rxe7+! Nxe7 21.Qxd7+! forces the king onto the mating square; 22.Bxe7# checkmate. Two rook sacs, one queen sac, mate in three. Steinitz called this the 'Evergreen' — the game that never fades from the beginner canon.
Morphy's Opera Game — Rxd7 rook sacrifice + mate
Morphy vs Duke of Brunswick + Count Isouard, Paris Opera 1858. 16.Qb8+! Nxb8 17.Rd8# — Morphy sacrifices the queen to bring the rook to the mating square with tempo. The most-played demonstration game in beginner instruction because it teaches development, king safety, and the deflection sacrifice in one 17-move package.
Tal's exchange sacrifice — game 6 vs Botvinnik 1960
Tal vs Botvinnik, World Championship match game 6, Moscow 1960. Tal plays Nf5 in a critical position, then when the knight is attacked, sacrifices it for two pawns and a decisive kingside attack. Tal's book The Life and Games of Mikhail Tal describes his internal reasoning: 'You can only calculate two moves. After that, you have to believe.' Tal won the match 12½-8½ using this style, becoming World Champion at 23.
Kasparov's positional Rxd7 — game 20 vs Kramnik 2000
Kasparov vs Kramnik, London 2000 World Championship match game 2 (the Berlin defence). Kasparov's exchange sacrifice for a structural attack that Kramnik defended perfectly. Modern annotators cite this game as the definitive example of an exchange sacrifice that is objectively sound but practically insufficient at the master level — Kramnik's Berlin drew the match and Kasparov lost his crown.
Deep Blue's Nb5 — game 4 vs Kasparov 1997
Deep Blue vs Kasparov, New York 1997 match. Move 8: Deep Blue plays Nb5!, a positional sacrifice-threat that at the time was considered inhuman because it wasted a tempo to prevent Kasparov's plan. Kasparov famously accused the team of human intervention because 'no computer would play this move'. The move was a pure positional sacrifice of tempo for permanent piece coordination — a category no engine before Deep Blue had reliably evaluated.
Carlsen's Rxb6 — round 8 Tata Steel 2015
Carlsen vs Radjabov, Tata Steel 2015 round 8. Carlsen sacrifices an exchange (Rxb6) to gain a passed pawn on the b-file that eventually queened. Modern classic exchange sacrifice — Carlsen calculated 15 moves ahead and traded 5 points for 3 points now, plus an inevitable queen 20 moves later. The engine confirmed +2.4 at the moment of sacrifice.
Sacrifice and Tempo — The Economics of a Gambit
Every gambit is a sacrifice priced in tempo: the gambit player accepts a material deficit for a specific number of extra developing moves. The market rate is roughly:
- Smith-Morra Gambit — one pawn buys two developing tempi.
- King's Gambit — one pawn buys a half-open f-file, central majority, and one attacking tempo.
- Danish Gambit — two pawns buy three developing tempi and the initiative.
- Evans Gambit — one pawn (b4) buys a tempo against Bc5 and the big centre.
Engines rate most gambits between +0.5 and +1.5 for the defender at master level — because at 2700, the defender finds the antidote. At club level, the same gambits are close to 50/50 because the defender consumes their hard-won tempi on finding the antidote rather than defending it.
Common Beginner Mistakes
The five mistakes that turn every legal capture into a losing gamble — and the fixes:
Sacrificing without a concrete follow-up
Every sacrifice must answer the question: 'what specifically happens after they take?' Name the next 3-5 moves. If you cannot see them, do not sacrifice. Spielmann's rule from The Art of Sacrifice (1935): the burden of proof is on the sacrificer — if you cannot demonstrate compensation, it's not a sacrifice, it's a blunder. The engine agrees: any move that drops material without a forcing sequence in the next 5-8 plies gets an evaluation drop of at least -1.0.
Confusing 'flashy' with 'sound'
A sacrifice that gives up a queen for three pieces and mate looks brilliant but is often objectively refutable. The engine evaluates by material + activity + king safety, not by dramatic effect. Before every sacrifice, ask: 'is this the SIMPLEST way to convert my position?' If a quiet strengthening move gets +2.0 and the flashy sacrifice gets +1.5, the quiet move wins. Save the sacrifices for positions where they're the ONLY way to make progress.
Sacrificing when you're not attacking
Attacking sacrifices work only when your pieces are aimed at the enemy king. If your queen is on a1, your rooks are on b1 and c1, and you sacrifice on h7, the queen and rooks cannot arrive at the h-file in time. Before every attacking sac, count: how many of your pieces can join the attack within 3 moves? Two or fewer = insufficient, refuse the sacrifice.
Not accepting a bad sacrifice
The reverse mistake: declining a sacrifice that you should take because the position 'looks scary'. If the engine says accepting the sacrifice is +2.0 for you, decline is -0.5, and the mating attack ends after 6 moves — take the material. Fear of the opponent's plan makes players decline free pieces at every rating level below 2200. The rule: if you cannot see the concrete winning line for the SACRIFICING side, take the material.
Treating every gambit as a sacrifice
A gambit is a specific opening sacrifice of a pawn for lead in development. A sacrifice is any deliberate offer of material for a bigger dividend at any point in the game. All gambits are sacrifices, but not all sacrifices are gambits. Middlegame exchange sacrifices, defensive sacrifices, and clearance sacrifices are not gambits. Reading Spielmann's typology once is enough to internalise the difference forever.
Practise Sacrifices for Free
Sacrifice is the concept that unlocks tactical vision — but it only becomes automatic when you calculate one every day. Three free, no-signup ways to train:
- Free PGN Viewer — load Anderssen's Immortal, the Evergreen, Morphy's Opera Game, and Tal vs Botvinnik 1960 game 6, and step through the calculation move by move.
- Free Tactics Trainer — unlimited sacrifice puzzles (queen sacs, exchange sacs, deflection sacs, mating combinations). Calibrate to your rating and solve 5 per day.
- chess.rodeo — no account, no paywall — play live games and let Stockfish flag every unsound sacrifice in the post-game review. Every “brilliant” idea that dropped material without compensation gets a red evaluation drop.
Frequently asked questions
Everything beginners and intermediate players ask about sacrifices — what qualifies as a sacrifice, Spielmann's real vs sham distinction, sacrifice vs gambit, queen sacrifices, exchange sacrifices, positional sacrifices, the Greek Gift, whether sacrifices are still sound in the engine era, and where to practise for free.
What is a sacrifice in chess?
A chess sacrifice is the deliberate voluntary offering of material — a pawn, minor piece, rook, or queen — to gain a bigger dividend such as tempo, initiative, structural damage to the enemy pawn shield, king exposure, or a long-term positional asset like a passed pawn or a permanent outpost. The distinguishing feature is intent: an accidental piece loss is a blunder, but the same move offered voluntarily with a plan for compensation is a sacrifice. The dividend can be immediate (mate in 6) or long-term (a passer that queens 25 moves later). Chess writer Rudolf Spielmann's 1935 book The Art of Sacrifice in Chess remains the definitive typology.
What is the difference between a real sacrifice and a sham sacrifice?
Rudolf Spielmann introduced the distinction in 1935. A REAL sacrifice permanently gives up material — the sacrificer accepts the deficit forever and lives off structural, positional, or psychological compensation. A SHAM sacrifice is a temporary offer that recovers all material (and often more) within a fixed forcing sequence, usually 2-8 moves. Anderssen's Immortal Game queen sac is a sham sacrifice because the mate follows within 3 moves; Petrosian's positional exchange sacrifices are real sacrifices because the material never comes back and the position never becomes tactical. Real sacrifices require positional judgement; sham sacrifices require concrete calculation.
What is the difference between a sacrifice and a gambit?
A gambit is a specific type of opening sacrifice — usually a single pawn — offered in the first 5-10 moves for lead in development, open lines, or initiative. Every gambit is a sacrifice, but not every sacrifice is a gambit. Middlegame piece sacrifices, exchange sacrifices, defensive sacrifices, and endgame study-style sacrifices are all sacrifices without being gambits. The King's Gambit is both a gambit AND a sacrifice; Petrosian's move-25 exchange sacrifice for a permanent dark-square grip is a sacrifice but not a gambit.
What is a queen sacrifice?
The offering of the queen — the strongest piece worth 9 points — for a smaller amount of material plus a decisive positional or mating dividend. Queen sacrifices are almost always sham sacrifices because the compensation must be enormous (usually forced mate or a queen recovery via promotion). Famous examples: Anderssen's Qxg7+ in the Immortal Game (1851), Morphy's Qb8+ in the Opera Game (1858), Nezhmetdinov vs Chernikov 1962 (18.Rxb7! + eventual queen sac), Reshevsky vs Fischer 1961 (Fischer's Nxb2). At club level, the queen sacrifice is the dopamine hit of chess — but only calculate to mate before you play one.
What is an exchange sacrifice?
Trading a rook (5 points) for a minor piece (3 points) — accepting a 2-point material deficit for a specific dividend. The most common types: (1) sacrifice on c3 or c6 in the Sicilian and King's Indian to damage the pawn structure and clear the file; (2) sacrifice on d5 or e5 to install a permanent knight outpost; (3) sacrifice on f6 or f3 to shatter the king shield. Engines now endorse positional exchange sacrifices more than any previous generation of theory — the Sicilian Najdorf Rxc3 and King's Indian Rxc3 lines are both engine-approved at master level.
What is a positional sacrifice?
A sacrifice where the material NEVER comes back and the position NEVER becomes tactical. Instead, the sacrificing side lives off a permanent structural asset: a passed pawn that cannot be stopped, a knight outpost the defender can never dislodge, a light- or dark-square weakness, or the bishop pair in an open position. Petrosian's exchange sacrifices, Karpov's strategic pawn sacrifices, and modern engine-approved sacrifices for compensation over 20+ moves are all positional. Positional sacrifices are the hardest to play because there is no tactical checkpoint that validates them.
What is the Greek Gift sacrifice?
The classic Bxh7+ (or Bxh2+) piece sacrifice on the h7 square in front of the castled king. The full sequence is: 1.Bxh7+ Kxh7 2.Ng5+ Kg8 (or Kg6) 3.Qh5 with mate on h7 next move. The bishop is sacrificed to force the king to h7 and open the h-file; the knight lands on g5 with tempo (check); the queen swings to h5 with a mating threat. The sacrifice is sound only when specific conditions are met (dark-squared bishop can't defend, no defensive Nf6, queen has access to the h5-h7 corridor). Named 'Greek Gift' after the Trojan Horse — apparently a gift, actually a trap.
What did Spielmann say about sacrifices?
Rudolf Spielmann (1883-1942), Austrian grandmaster and one of the most brilliant attacking players of his era, wrote The Art of Sacrifice in Chess in 1935 — still the definitive typology today. His key contributions: (1) the real-vs-sham distinction (real sac = material never returns; sham sac = temporary offer with concrete recovery); (2) the classification of sacrifices by dividend (mating attack, development, positional, defensive, obstructive); (3) the rule that the burden of proof is on the sacrificer — you must be able to name the compensation before you make the move. Every subsequent sacrifice-instruction book (Vukovic, Nunn, Aagaard) builds on Spielmann's framework.
Are sacrifices sound in modern engine chess?
Yes — often more than pre-engine theory suggested. Modern engines (Stockfish, Leela) now endorse many positional exchange sacrifices, some pawn sacrifices for space, and specific piece sacrifices for king-safety attacks that classical writers rejected. What engines refute is the FLASHY unsound sacrifice — a queen sac that leads to +0.5 evaluation instead of +2.0 with a quiet move gets rejected. The engine era rehabilitated the positional sacrifice (Petrosian's technique) while retiring the speculative attacking sacrifice that only worked because the defender couldn't find the antidote. At club level, most Romantic-era sacrifices still work because the defence remains unknown territory.
When should I sacrifice material in my own games?
Three conditions must all be true: (1) you have a concrete calculated follow-up naming at least 3 moves after the sacrifice; (2) at least half your pieces can join the attack or claim the dividend within 3 moves; (3) the quietest strengthening alternative does NOT give a bigger evaluation. If any condition fails, don't sacrifice. At club level, most 'brilliant' sacrifices fail condition 1 (no concrete follow-up) or condition 3 (the quiet move was better). The habit that separates 1500 from 1900: pause before every sacrifice and ask 'is there a boring move that's also winning?' If yes, play the boring move.
What is the hardest kind of sacrifice to calculate?
The positional sacrifice for long-term compensation — because there is no calculable end. An attacking sacrifice ends in mate or a material recovery within 5-10 moves; you can verify the compensation with concrete calculation. A positional sacrifice pays off 20-40 moves later as a converted structural advantage that no calculation can guarantee. Petrosian's exchange sacrifices are the archetype: the compensation is a light-square grip, a passed pawn, or an outpost that will slowly constrict the defender over the entire second half of the game. Only strong intuition and years of positional pattern-matching validate a positional sacrifice.
What are the most famous sacrifices in chess history?
Anderssen's Immortal Game (1851, two rooks + queen); Anderssen's Evergreen (1852, queen sac Qxd7+); Morphy's Opera Game (1858, Qb8+ Nxb8 Rd8#); Steinitz vs von Bardeleben 1895 (11-move mating attack after Rxe7); Rotlewi vs Rubinstein 1907 (double bishop sacrifice with mate); Nezhmetdinov vs Chernikov 1962 (queen sac + mating attack); Tal vs Botvinnik game 6 1960; Kasparov vs Topalov 1999 Wijk aan Zee (Rxd4+); Kasparov vs Karpov game 16 1985 (Rxc3); and Carlsen's exchange sacrifices in the 2010s. Every strong player learns this canon before rating 1800.
What is a defensive sacrifice?
Returning material — often exactly the amount you're up — to blunt an incoming attack, trade queens off the board, or force a drawn endgame. The classic pattern: you win a piece early, the opponent starts a mating attack against your king, you give back the piece to eliminate the strongest attacker or block the key attacking line. Karpov's 1974 exchange-back-sacrifice against Korchnoi (returning a whole exchange to reach a technically drawn rook ending) is the most-cited example. Modern engines endorse defensive sacrifices frequently — giving back material to reach an opposite-coloured-bishop endgame or a fortress is a standard tool at every rating level.
Where can I practise chess sacrifices for free?
Load master sacrifice games in our free PGN Viewer — Anderssen's Immortal, Morphy's Opera Game, Tal-Botvinnik 1960 game 6, Kasparov-Topalov 1999 — and step through the calculation move by move. Use our free Tactics Trainer for unlimited sacrifice puzzles (queen sacs, exchange sacs, mating combinations). Then play live games at chess.rodeo where Stockfish flags every unsound sacrifice in the post-game review with a red evaluation drop — every 'brilliant' idea that dropped material without compensation gets an eval report. No account, no signup, no paywall on any of the three tools.
Play any game — let Stockfish flag every unsound sacrifice free
Import your games into chess.rodeo, hit Analyze, and every 'brilliant' sac that dropped material without compensation gets a red evaluation drop. Free, no account, no paywall.