Bishop against knight · For club players · 19 min read
Bishop vs Knight: Fischer's Famous Exchange, Checked by Stockfish
In game 7 of the 1971 Candidates final, Fischer traded a dominant knight for a bad bishop and won a textbook bishop-against-knight ending. Stockfish 19 says the trade was not best. Here is why it worked anyway.
- plan
- target
- Knight's reach
- pawn chain or line of force
In short
- A bishop beats a knight when there are pawns on both wings and the enemy pawns are fixed on the bishop's colour: it reaches both wings in one move, and the knight needs three.
- Fischer built that ending on purpose: 18.b4 fixed Black's a6 pawn on a light square, 19.Nc5 occupied the outpost in front of it, and 20.f3 took the e4 and g4 squares from the knight.
- The famous 22.Nxd7+ traded White's dominant knight for Black's bad bishop. Stockfish 19 says keeping the knight with 22.a4 was stronger by more than a pawn.
- Petrosian could still have held with active moves (23...d4, 27...Nb6, 28...Nb6 or 28...d4). The losing move was 28...f5, which put a fourth pawn on the bishop's colour and opened the seventh rank.
- To use the plan: fix the enemy pawns on your bishop's colour, take the outposts from the knight, put a rook on the seventh, and bring the king to the centre.
How we know what we say
- Verified From the record: Stockfish's numbers, the players' own notes or interviews, or a cited source.
- Inferred Read from the moves and the position: a reasoned reconstruction, not the player's words.
- Speculative A guess about mood, intent or practical motives: plausible, not proven.
Part 1The setting: a winning streak and a former champion
In 1971 Bobby Fischer played three matches to earn the right to challenge the world champion. Fischer beat Mark Taimanov 6–0 and Bent Larsen 6–0, part of a run of 20 straight wins (one of them by forfeit) that Tigran Petrosian ended in game 2 of their final in Buenos Aires. [7] Verified
Petrosian was the hardest opponent left. World champion from 1963 to 1969, Petrosian was famous for prophylaxis: seeing the opponent's plan first and quietly stopping it. Beating Petrosian meant winning games where nothing obvious happens, which is exactly what game 7 is. The match stood at 3½–2½ to Fischer when it began. [5] Verified
| Game | White | Result | Match score |
|---|---|---|---|
| 1 | Fischer | 1–0 | 1–0 |
| 2 | Petrosian | 1–0 | 1–1 |
| 3 to 5 | alternating | three draws | 2½–2½ |
| 6 | Petrosian | 0–1 | 3½–2½ |
| 7 | Fischer | 1–0 | 4½–2½ |
| 8 | Petrosian | 0–1 | 5½–2½ |
| 9 | Fischer | 1–0 | 6½–2½ |
Game 6 had been a long fight that Fischer won after an adjournment, and later analysis found that Petrosian's sealed move threw away a drawn position. Game 7 was the next game. [6] Verified
Part 2The idea at a glance: why a bishop outruns a knight
A bishop and a knight are both worth about three pawns, but they are good at different things. A bishop moves any distance along its diagonals, so in an open position it can attack on the queenside and the kingside in the same move. A knight hops a short, fixed distance: to cross from one wing to the other it needs three or four moves. It is strongest on a protected square in the middle of a closed position.
- target
- key square
- pawn chain or line of force
- outpost
- target
- pawn chain or line of force
The same two pieces, opposite verdicts. The bishop wants open lines and targets on its colour; the knight wants blocked pawns and a permanent outpost. Fischer spent twenty moves making the game look like the left-hand picture, and the numbers show why that takes patience.
Part 3The opening: Fischer's quiet path to two weaknesses
Petrosian chose the Kan Sicilian (the Openings section covers the Sicilian's main lines), and Fischer answered with the plain 5.Bd3 and 6.Nxc6, inviting Black to recapture towards the centre with the b-pawn. When Black struck back with 7...d5 and 8...Nf6, Fischer opened the centre, and after 12.Qa4+ Qd7 13.Re1 Qxa4 14.Nxa4 the queens were off.
1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 a6 5.Bd3 Nc6 6.Nxc6 bxc6 7.O-O d5 8.c4 Nf6 9.cxd5 cxd5 10.exd5 exd5 11.Nc3 Be7 12.Qa4+ Qd7 13.Re1 Qxa4 14.Nxa4 +0.88White slightly better
The opening as played. The eval at the end is Stockfish's verdict on the queenless position: a small but lasting edge for White. Verified
- plan
- outpost
- target
With the queens gone, the game is about structure, not attack. The verdict that matters is not the small engine edge but what it is made of: two fixed weaknesses and an outpost, none of which can ever go away. Inferred
Nothing in the opening was a clear error by Stockfish's measure: every move of both players stays within about half a pawn of its first choice. The damage was structural, and it came from moves that looked natural: 6...bxc6 and the exchanges on d5 left Black with the two isolated pawns. Verified
Part 4The plans: squeeze against counterplay
From move 15 to move 20 both sides set up for the ending. Fischer traded the dark-squared bishops with 15.Be3 and 16.Bc5, played 18.b4 so that ...a5 can be met by bxa5, and put the knight on c5 with 19.Nc5. Then 20.f3 took away e4 and g4, the two squares the black knight most wanted.
- plan
- target
- outpost
- White's grip
- plan
- key square
- Black's route
One position, two plans. White's plan is slow and permanent; Black's depends on activity, and every passive move makes it harder. That trade-off decides the game.
Part 5The critical moments
Three decisions turned a slightly better ending into a winning one, and Stockfish's view of them is not the one in most books. Try each position before opening the answers.
Critical moment 1 of 5 · Move 20, Black to play
Challenge the knight, or wait?
- key square
- target
White's knight on c5 dominates the queenside. Black has a choice between challenging it now and tidying up the rooks. What would you play?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 20...Nd7 21.Rec1 Re3 | +0.19about equal | Stockfish's choice. The knight goes to d7 to challenge c5 at once: if White's knight leaves or is exchanged, Black's weaknesses matter much less. Verified |
| 20...Ne8 | +1.06White clearly better | Heads for c7 or d6 to cover the queenside, but it is slow: Stockfish rates it almost a pawn worse than Nd7, close to the game move. Verified |
| 20...Rxe1+ 21.Rxe1 | +0.72White slightly better | Trades a pair of rooks and gives the remaining White rook the e-file. Stockfish's second choice here: about half a pawn worse than Nd7, but better than the game move. Verified |
What Petrosian played
20...Rea7+1.36White clearly better
Petrosian doubled the rooks on the a-file behind the pawn. It protects a6 but puts both rooks on a passive file and leaves the knight on c5 unchallenged.
Stockfish preferred 20...Nd7 (+0.17); the game move gave up about 1.2 pawns. Depth 27.
What it revealed
Petrosian chose solidity over activity, the defender's instinct of guarding a weakness rather than removing the piece that attacks it. Against a slow squeeze, that instinct is often wrong: the guarding pieces stop doing anything else.
- Verified Stockfish rates 20...Nd7 as clearly better than the game move: the passive rook move costs Black more than a pawn.
- Inferred Doubling rooks on the a-file only makes sense as pure defence of a6; nothing in the position suggests a plan of ...a5, which b4 prevents.
- Speculative Petrosian may have trusted the position to hold without risk, as the safety-first style that had made Petrosian world champion often did.
Critical moment 2 of 5 · Move 22, White to play
Keep the dominant knight, or trade it?
- outpost
- target
After 21...Bd7, Black's bishop offers itself on d7 and White's knight on c5 attacks it. Do you keep the knight, the best piece on the board, or trade it?
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 22.a4 22...Bc6 23.Ree1 | +2.32White winning | Stockfish's choice. Keep the dominant knight against Black's bad bishop, which is hemmed in by its own pawns on a6 and d5, and prepare a5 or b5 to fix the queenside for good. Verified |
| 22.Kf2 | +0.37White slightly better | The natural improving move, but it gives Black the tempo to untangle with Bb5 or Nd7: Stockfish rates it below even the game move. Verified |
| 22.Rc1 | +0.32about equal | Takes the c-file at once, but Black answers with Bb5 or Bc6 and the bishop improves: Stockfish rates it below the game move too. Verified |
What Fischer played
22.Nxd7++0.93White clearly better
Fischer took the bishop with check. It trades a good knight for a bad bishop, which breaks a textbook rule, to reach an ending of bishop against knight with pawns on both wings, where White's bishop has targets on a6 and d5 and Black's knight has no outpost.
Stockfish preferred 22.a4 (+2.04); the game move gave up about 1.1 pawns. Depth 27.
What it revealed
Fischer valued the clarity of a pure, technical ending over the bigger but messier advantage of the knight. It is a classic transformation: one advantage, the knight against a bad bishop, swapped for another, a good bishop against a knight with no squares.
What Stockfish adds Stockfish 19 rates 22.a4 at more than two pawns for White and the game move at under one: the famous exchange gave back more than a pawn of the advantage. But of the natural moves, only 22.a4 does better; the quiet 22.Kf2 and 22.Rc1 both let Black untangle. The exchange won because of what Black did next, not because it was forced.
- Verified Karsten Müller put exactly this question to ChessBase readers in 2020: was 22.Nxd7+ objectively best, or was 22.a4 stronger?[3]
- Verified In the published solution, Zoran Petronijević concluded that 22.Nxd7+ objectively leads only to a draw, and that most engines prefer 22.a4.[4]
- Inferred The exchange removes Black's only piece that could cover the light squares, so every black pawn on a light square becomes a permanent target of the bishop.
- plan
- idea not played
- outpost
- target
- plan
- target
- pawn chain or line of force
Compare the engine's numbers above each board. Stockfish keeps the knight; Fischer chose the ending a human can play almost without risk.
Critical moment 3 of 5 · Move 23, Black to play
The pawn move that frees the knight
- threat
- key square
- target
White's rooks are coming to c7 and e7. Black's knight on f6 has no good square. Find the move that gives it one.
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 23...d4 24.a3 Nd5 25.Kf2 Rad8 | +0.87White slightly better | Stockfish's choice. The pawn steps off the bishop's colour and gives up d5 to the knight, which heads for c3. Black's pieces come alive and White's edge stays under a pawn. Verified |
| 23...Rc8 24.Rxc8+ Rd8 25.Rxd8+ Ne8 26.Rexe8# | #3White mates by force | Contesting the file with the other rook looks natural but loses at once: 24.Rxc8+ Rd8 25.Rxd8+ Ne8 26.Rexe8 is mate. Black's back rank is as weak as the pawns. Verified |
| 23...Ne8 | +1.54White clearly better | Guards c7 from e8, but the knight is even more passive there and the d5 pawn stays a target. Verified |
What Petrosian played
23...Rd6+1.70White clearly better
Petrosian brought the rook to d6 to defend along the sixth rank. It covers a6 and d5 but leaves the knight on f6 without a future and lets White's rook land on c7.
Stockfish preferred 23...d4 (+0.77); the game move gave up about 0.9 pawns. Depth 30.
What it revealed
Again Petrosian guarded the weaknesses instead of changing the structure. The d-pawn was Black's only way to escape the bishop's colour, and staying put kept every target in place.
What Stockfish adds With 23...d4 Stockfish has White only slightly better; after 23...Rd6 24.Rc7 the advantage is back to about one and a half pawns.
- Verified Petronijević named 23...d4 as Black's best defence and 23...Rd6 as the mistake that leads to a lost position; Charles Sullivan's lines show the position after 23...d4 is about level.[4]
- Inferred A pawn on d4 is on a dark square, out of the bishop's reach, and it vacates d5 for the knight: two problems solved with one move.
Part 6The turning point: 28...f5
By move 28 Fischer's rook sat on c7 and the king was marching up the board, but the engine says Fischer had not been perfect either: 27.f4 and 28.Kf3 each let Black back towards equality. The position below is one where Black can still hold.
Critical moment 4 of 5 · Move 28, Black to play
One pawn move too many
- key square
- target
- Seventh rank
Black is under pressure but not lost. White's king is on f3 and the rook is on c7. Find the move that keeps Black's pawns off the bishop's colour and the knight active.
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 28...Nb6 29.g4 hxg3 30.hxg3 d4 | +0.09about equal | One of Stockfish's two choices. The knight heads for c4 or d5, and with ...d4 to follow Black is about level. Verified |
| 28...d4 | +0.21about equal | Stockfish's other choice: the pawn leaves the light squares, the knight gets d5 and e5, and Black is about level. Verified |
| 28...Rd8 | +0.20about equal | Passive but still playable: Stockfish rates it about as good as d4. Verified |
What Petrosian played
28...f5+2.07White winning
Petrosian pushed the f-pawn, most likely to keep White's king out of e4 and to fix White's f4 pawn. It also put a fourth black pawn on a light square and took the f-pawn off f7, so the seventh rank opened for White's rooks.
Stockfish preferred 28...d4 (+0.11); the game move gave up about 2.0 pawns. Depth 23.
What it revealed
Under pressure, the defender reached for a move that stops one thing, the king, and missed what it gave away: the light squares and the seventh rank. This is how quiet positions are usually lost.
What Stockfish adds Stockfish rates 28...f5 as the decisive mistake: the position goes from about level to more than two pawns for White. From here the engine never lets White's advantage drop.
- Verified Petronijević also called 28...f5 Black's decisive error, and found draws with 27...Nb6 and with 28...d4.[4]
- Inferred After 28...f5 Black's pawns stand on a6, d5, f5 and g6, all light squares, and only h4 is on a dark one: every pawn but one is now a target for the bishop.
- Speculative After the long game 6 and a long passive defence here, fatigue may have played a part; nothing in the record says so directly.
- the move
- target
- The seventh rank
29.Ke3 d4+ 30.Kd2 Nb6 31.Ree7 Nd5 32.Rf7+ Ke8 33.Rb7 Nxf4 34.Bc4 +10.26White completely winning
The rest of the game. Both rooks reach the seventh, the king steps to d2 in front of Black's d-pawn, and the bishop finishes the job. Verified
Part 7What Petrosian could have done
Black's defence failed for one reason, repeated four times: every time the choice was between guarding a weakness and changing the structure, Petrosian guarded. Stockfish finds a save at almost every step, and they share a theme.
- plan
- Knight's new squares
Critical moment 5 of 5 · Move 27, Black to play
The last clear chance
- target
- key square
Fischer has just played 27.f4. The knight on d7 has been passive for three moves. Find its best square.
Think it through on the boardWeigh the candidates
| Move | Stockfish | Verdict |
|---|---|---|
| 27...Nb6 28.f5 Rf6 29.Ke3 gxf5 30.Kd4 Nc4 | +0.28about equal | Stockfish's choice. The knight reaches c4, an outpost of its own, and Black is only slightly worse. Verified |
| 27...d4 | +0.72White slightly better | The same idea of getting off the light squares and freeing d5, but less precise: Stockfish leaves White about half a pawn more than after Nb6. Verified |
| 27...Rb8 | +0.48White slightly better | Active and playable, between Nb6 and d4 by Stockfish's count: the rook attacks the b4 pawn down the b-file. Verified |
What Petrosian played
27...h4+1.90White clearly better
Petrosian pushed the h-pawn to gain space on the kingside. It does not lose by force, but it spends a tempo away from the queenside, where the game is decided, and leaves the knight on d7.
Stockfish preferred 27...Nb6 (+0.19); the game move gave up about 1.7 pawns. Depth 32.
What it revealed
Petrosian kept looking for a safe pawn move rather than an active piece move. In an ending where the opponent has the better minor piece, tempo spent on pawn moves usually helps the side with the better piece.
What Stockfish adds Fischer's 27.f4 was itself imprecise: Stockfish prefers 27.a4, keeping control of b5, and after 27.f4 the best defence, 27...Nb6, brings Black back towards equality.
- Verified Petronijević called 27...h4 a mistake and found that 27...Nb6, recommended by almost every annotator, saves the game.[4]
- Inferred No white pawn can ever attack c4: the b-pawn is already on b4 and White has no d-pawn, so only White's pieces could challenge a knight there.
27...Nb6 28.f5 Rf6 29.Ke3 gxf5 30.Kd4 Nc4 +0.72White slightly better
Stockfish's defence after 27.f4: the knight comes to c4 and Black's pieces finally work together. Verified
Part 8A different angle: the famous move was not where the game was won
Most accounts of this game crown 22.Nxd7+ as the winning idea. Checked move by move with Stockfish 19, the story looks different. White's advantage was built by Black's passive 20th and 21st moves, given back in part by the famous exchange and by two imprecise moves later, and decided by Black's 28th move. Verified
| Move | Played by | Stockfish's choice | What it cost |
|---|---|---|---|
| 20...Rea7 | Petrosian | 20...Nd7 | more than a pawn |
| 21...Bd7 | Petrosian | 21...g6 | about a pawn |
| 22.Nxd7+ | Fischer | 22.a4 | more than a pawn |
| 23...Rd6 | Petrosian | 23...d4 | almost a pawn |
| 27.f4 | Fischer | 27.a4 | a pawn and a half |
| 27...h4 | Petrosian | 27...Nb6 | more than a pawn and a half |
| 28.Kf3 | Fischer | 28.Rec2 | almost two pawns |
| 28...f5 | Petrosian | 28...Nb6 or 28...d4 | about two pawns |
There is a lesson in the reverse direction too. The engine says Fischer erred on moves 27 and 28, and both times Petrosian missed the chance. Pressure does not need to be perfect to work; it needs to keep the defender making hard decisions. Inferred
One more caution. Engines judge positions, not people. A choice that is worse by the engine's count can still be the better choice at the board if it leaves the opponent more ways to go wrong. What the engine does settle is that Black was never lost before move 28. Inferred
Part 9How to use this in your own games
You do not need Fischer's technique to use Fischer's plan. The bishop-against-knight ending is one of the most common in club chess, and its recipe is short.
The bishop's recipe
- Check the pawnsAre there pawns on both wings, and can the position open? If every pawn is blocked, or all the pawns are on one wing, the knight is usually at least as good.
- Fix the targets on the bishop's colourMake the enemy pawns stop on your bishop's colour, the way 18.b4 fixed a6. A fixed pawn on the right colour is a target forever.
- Take the squares from the knightFind the outposts the knight wants and cover them with pawns, as 20.f3 took e4 and g4. A knight without a safe square is just a slow piece.
- Use the files and the seventh rankRooks become stronger as the position opens. Aim one at the seventh rank, where it attacks pawns from the side and confines the king to the back rank.
- Bring the king to the centreWith the queens off, the king is a fighting piece. Fischer's king walked from g1 to d2 while the rooks worked.
- Trade the right piecesKeep the piece that matters most, and trade the defender's best piece. Fischer gave up the knight because the ending without it was clearer; be ready to keep it when the knight is the stronger piece.
- Before each trade, ask which minor piece the ending suits, not which piece is better right now.
- As the defender, move pawns off the enemy bishop's colour early, while it is still possible.
- As the defender, give the knight an outpost before the attacker takes the squares away.
- Do not answer a slow squeeze with slow moves: activity is the defender's only currency.
Your turnPractise it
Set each position up, decide your plan before you look, then play it out against Stockfish to see whether it holds.
White to play after 22...Rxd7: how do White's rooks get into Black's position?
Hint
Which file is open, and which rank would hurt Black most?
A good answer
23.Rc1, then Rc7: the c-file is open and the seventh rank is where Black's pawns and king are. Stockfish's first choice is 23.Rc1, as in the game.
White to play after 27...h4: Fischer played 28.Kf3. Find the move that keeps White's advantage largest.
Hint
Double on a file before the king walks; a rook trade can create a passed pawn.
A good answer
28.Rec2. After 28...Rb8 29.Rc8+ Kg7 30.Rxb8 Nxb8 31.b5 axb5 32.Bxb5 White's a-pawn is passed and the bishop is free. Stockfish rates this more than a pawn better than the game move.
White to play after 29...d4+: where should the king go?
Hint
There are three legal king moves. Think about the d-pawn and the e-file.
A good answer
30.Kd2, as Fischer played. The king stands in front of the d-pawn and stops it on d3, stays near the queenside, and leaves the rook on e2 free to reach e7. By Stockfish's count either other king move throws away almost all of White's advantage.
White to play after 31...Nd5: both rooks are on the seventh rank. How do they make it count?
Hint
A check first, then look at the queenside.
A good answer
32.Rf7+ drives the king to e8, and 33.Rb7 swings the second rook across to attack along the seventh rank and behind the queenside pawns. Both are Stockfish's first choices, and both were Fischer's moves.
Part 10What to take from this game
- Structure first. The game was decided by two weak pawns that appeared on move 14 and never moved again.
- Fix, restrict, invade. Fischer fixed the pawns, took the knight's squares, and only then put the rooks on the seventh rank.
- Transform, but count. Swapping one advantage for another is a powerful technique; Stockfish shows it can also cost something. Know what you are giving up.
- Defend actively. Petrosian's saving moves were all active: a pawn off the bishop's colour, a knight to an outpost.
- Check the famous moves. Even the most celebrated moves in chess history can be improved, and the improvements are where the learning is.
Questions readers ask
Is a bishop better than a knight in the endgame?
Not always. A bishop is usually better when there are pawns on both wings and the position is open, because it reaches both sides quickly. A knight is better in blocked positions, with all the pawns on one wing, or when it has a protected outpost that no pawn can attack.
Was Fischer's 22.Nxd7+ against Petrosian the best move?
By Stockfish 19's evaluation, no: 22.a4, keeping the knight on c5, kept an advantage of more than two pawns, while 22.Nxd7+ left about one. Published analysis, including Zoran Petronijević's for ChessBase, reached the same conclusion. The exchange still worked because Black did not find the active defence.
Why did Petrosian lose this endgame?
Petrosian repeatedly chose passive defence over changing the structure. Active moves such as 23...d4, 27...Nb6 and 28...Nb6 or 28...d4 would have held, according to Stockfish and published analysis. The decisive mistake was 28...f5, which fixed another pawn on the bishop's colour and opened the seventh rank for White's rooks.
Where should I put my pawns against an enemy bishop?
On the other colour from the bishop. Pawns on the bishop's colour can be attacked by it and can only be defended by your own pieces; pawns on the other colour are out of its reach and also restrict it.
When should I trade my knight for the opponent's bishop?
When the ending that remains suits your bishop better than your knight: pawns on both wings, enemy pawns fixed on your bishop's colour, and no outpost for the enemy knight. Keep the knight if it has a permanent outpost or the opponent's bishop is the bad one, as Stockfish preferred here.
Sources and method
The game score was checked against two independent sources. Every position, line and candidate in this article was replayed for legality, and every engine number was produced by Stockfish 19 for this article (depth 20 to 40, deepest at the critical moments), not copied from elsewhere. What the players thought is reconstructed from the moves, their published notes and later analysis; it is labelled as verified, inferred or speculative, and nothing is presented as a quotation unless a source is cited for it.
- Fischer v Petrosian, Candidates final, game 7, Buenos Aires 1971: game score and PGN (chessgames.com)
- The same game score with the Lichess video "Amazing Game: Bobby Fischer vs Tigran Petrosian" (lichess.org)
- Karsten Müller, "Historical riddle: Was Fischer's 22.Nxd7 winning?", ChessBase, 18 November 2020
- Karsten Müller, the riddle's solution with analysis by Zoran Petronijević and Charles Sullivan, ChessBase (Spanish edition)
- Fischer v Petrosian Candidates final 1971: all nine games (chessgames.com)
- Karsten Müller, "Could Petrosian have held against Fischer?" (game 6), ChessBase
- Candidates Tournament 1971 (Wikipedia)
Read next
Kasparov's knight on d3 in game 16 of the 1985 match, what Stockfish says about it, and how to build an outpost no pawn can attack.
Bishop Pair Strategy: Kramnik v Tiviakov 2001, Checked by StockfishKramnik, the new world champion, turned the two bishops into a win at Wijk aan Zee 2001. Stockfish 19 shows what the pair was worth and where both players let the advantage slip.
Active King in the Endgame: Capablanca's King Walk, CheckedCapablanca gave up two pawns to march the king into Tartakower's camp in 1924. Stockfish 19 says the king walk was sound, but the game was decided two moves earlier.
Turn reading into skill
Strategy sticks when you use it. Practise real positions with Stockfish's verdict at once, or play the plans out from any diagram above.