Wrong Bishop

The wrong bishop is the bishop that does not control the promotion square of its own rook pawn. King, rook pawn, and wrong bishop against a lone king is a draw — the most famous fortress in endgame theory — provided the defending king reaches the corner in front of the pawn. An entire extra piece and pawn, and no win: chess geometry at its most merciless.

I call this the most profitable draw in chess, because it is decided long before the ending appears on the board. The player who knows the wrong-bishop rule steers pawn trades and piece trades twenty moves ahead of the player who does not — and wins or saves the same position the other player would lose.

The Exact Position

Here is the fortress at its purest. White: king on g6, pawn on h6, bishop on e4. Black: king on h8. White has an extra bishop and an extra pawn, the pawn is two steps from queening, and the position is a dead draw — I have verified it to the end with the engine, and White cannot make progress with any sequence of moves.

Why? Follow the colors. The pawn promotes on h8, a dark square. White's bishop on e4 is a light-squared bishop: it can attack every light square on the board and will never, until the end of time, touch h8. That is the entire problem. Black's king simply sits on h8 and g8, and White has no way to evict it:

The bishop cannot attack h8, and cannot attack g8 and h7 at the same moment in a useful way — checks just shuffle the king between its two home squares.
If White ever plays h7 with the black king on h8, the position is immediate stalemate: the king has no moves and is not in check.
White's king would love to control both g8 and h8 itself, but standing on g6 or h6-adjacent squares it can cover the exits only at the price of stalemate again. Every single winning attempt runs into the same wall.

Now flip one detail: give White a dark-squared bishop — say on f8 — and the game ends instantly: 1.Bg7+ Kg8 2.h7 is checkmate. Same material, same squares, opposite bishop color, and the difference is the whole point. The bishop that controls the promotion corner is the "right" bishop; the one that does not is "wrong."

The color logic is easy to reconstruct at the board, so you never need to memorize cases: the bishop is wrong whenever it does not control the pawn's promotion corner. For White, h8 is dark and a8 is light — so a light-squared bishop is wrong with the h-pawn, and a dark-squared bishop is wrong with the a-pawn. For Black, mirror it: h1 is light, a1 is dark.

The One Condition: The King Must Reach the Corner

The fortress has a strict entry requirement — the defending king must actually get to the corner (the h8/g8 zone in our example) before the attacker's forces seal it off. This is not a formality. Move the defending king far away — say to d3 in the same position — and White wins comfortably: the pawn runs, the bishop and king shepherd it, and the promotion happens long before the defender arrives. I verified this too: with the king cut off, it is a forced win in a handful of moves.

So every wrong-bishop ending reduces to a race you can evaluate with the same instincts as the square rule: can the defending king reach the corner zone in time? If yes, the game is a draw regardless of anything else on the board's empty half. If no, the "wrong" bishop wins as easily as a right one. When I calculate these races, I give the defender generous margins — the attacker's king and bishop can sometimes delay the trip with checks and shoulder-blocks — but the principle is binary, and so is the result.

One subtlety worth knowing: the defense is genuinely inexhaustible once the king is home. There is no waiting maneuver, no triangulation, no clever zugzwang that breaks it, because stalemate guards every door. This is a true fortress in the strict sense — the defender has safe moves forever, and the attacker's extra material is simply spectating.

How the Rule Shapes Earlier Play

Here is where the half points are actually earned. The wrong-bishop draw is rarely reached by accident; it is reached by one player aiming at it for many moves. The practical habits:

Defending a difficult bishop ending? Count what remains after trades. If you can give up your last pieces and pawns to leave the opponent with exactly the rook pawn plus wrong bishop, do it at almost any cost — a bishop sacrificed for the opponent's g-pawn, leaving only the h-pawn behind, converts a lost ending into a textbook draw on the spot. This bishop-for-pawn desperado is one of the most common saving resources in practical play, and spotting it early is the skill.

Winning a bishop ending? Guard your pawn inventory. The moment your winning plan depends on a rook pawn, check your bishop's color against the promotion corner. If they disagree, treat every trade of your other pawns as a potential half point leaking away. Often the correct technique is to keep a second pawn — any pawn on any other file — alive purely as insurance: with one extra non-rook pawn on the board, the fortress never forms.

Choosing which bishop survives. In middlegame trades where you hold the bishop pair, the endgame's corner colors are a real tiebreaker. If the likely long-term scenario is you pushing an h-pawn, the dark-squared bishop (for White) is the one that converts; if you are the one defending, maneuver so the opponent is left with the mismatched pair. Grandmasters make these color bookkeeping decisions routinely around move 25, which looks like clairvoyance and is actually just this rule.

Rook endings and everything else. The motif leaks into other endings: a defender down an exchange can sometimes return the rook for the last "good" pawn, banking on the wrong-bishop draw behind it. Whenever a rook pawn and an enemy bishop share the board, run the color check before evaluating anything.

Other Wrong-Corner Fortresses

The corner's color poisons more than the basic ending, and the family resemblances are worth knowing:

Doubled rook pawns do not help. Wrong bishop with two h-pawns is exactly as drawn as with one — the extra pawn adds nothing, since the problem was never material.
The attacker's king in the corner is a related disaster. Even in pure pawn endings, the rook pawn carries corner pathologies: an attacking king trapped on the promotion square in front of its own pawn, or a defender reaching the corner first, produces the standard rook-pawn draws. The edge of the board is systematically the defender's ally — the same geometry that makes the Lucena position fail for rook pawns.
Opposite-colored bishop blockades. With bishops of opposite colors, the defender plants the king and bishop on the color the attacker's bishop can never touch — a generalization of the same idea: winning attempts die where your bishop has no votes. A passed rook pawn whose queening square belongs to the defender's color complex is often unwinnable even with extra material elsewhere.

The unifying lesson I give students: a bishop is not half a rook — it is a full piece on one color and an absent piece on the other. Endings are where that absence stops being abstract and starts deciding results.

Training It

Two drills cement this pattern. First, hold the defense: set up our diagram position, let an engine attack with king, pawn, and wrong bishop, and feel how the g8/h8 shuffle plus stalemate holds against everything — most students need to experience the impossibility to trust it. Second, and more important, play the races: positions where the defending king sprints for the corner while the attacker tries to cut it off. Our endgame trainer includes the wrong-bishop fortress and its entry races against precise play, which is where the rule turns from trivia into saved half points. After that, you will find yourself checking corner colors in the middlegame — which is exactly the habit that wins these endings.

Frequently Asked Questions

What exactly makes a bishop the "wrong" bishop?

Color mismatch with the promotion corner of a rook pawn. If the bishop does not control the pawn's queening square (h8/a8 for White, h1/a1 for Black), it can never evict or checkmate a king camped there, and the defender draws by sitting in the corner. The rule involves only rook pawns — with any other pawn, the promotion square is not in a corner and the king can be driven off it.

Which combinations are wrong for White and for Black?

For White: h-pawn with a light-squared bishop, or a-pawn with a dark-squared bishop. For Black: h-pawn with a dark-squared bishop, or a-pawn with a light-squared bishop. Do not memorize the table — just look at the promotion corner's color and your bishop's color; if they differ, the bishop is wrong.

Does an extra pawn break the fortress?

An extra pawn on any other file does — the defender's king cannot guard the corner and stop a second passer elsewhere, and stalemate stops working once the attacker has spare pawn moves. But an extra doubled rook pawn does not help at all: two h-pawns with the wrong bishop are as drawn as one.

What if the defending king cannot reach the corner?

Then there is no fortress and the wrong bishop wins like any other piece: the pawn is escorted home, with the bishop covering the king's approach. The entire evaluation of these endings is the race to the corner — defender inside means draw, defender outside means loss — which is why you should start calculating that race the moment a wrong-bishop scenario appears on the horizon.

Why is it stalemate and not zugzwang that saves the defender?

Because the defender wants to run out of moves. In the corner, every attempt by the attacker to take away the king's last squares produces stalemate — an immediate draw — rather than zugzwang, which requires the victim to have bad moves available. The wrong-bishop fortress is built on the stalemate rule; in a chess variant where stalemate lost, the "wrong" bishop would win easily.

Related Terms

Fortress — the general concept; the wrong-bishop corner is its most famous example.
Stalemate — the rule that holds the fortress together.
Bad Bishop — a different bishop ailment: blocked by its own pawns rather than mismatched with a corner.
Bishop Pair — owning both colors, and the trade decisions where corner color becomes a tiebreaker.
Square Rule — the racing tool for judging whether the defending king reaches the corner.
Lucena Position — the rook-endgame win that likewise fails on rook pawns; the edge of the board defends.