Queen vs Rook
Queen versus rook is a forced win for the queen — at most thirty-five moves from the worst position, by tablebase count — and it is famously the hardest of the "winning" endgames to actually win. The defense keeps rook and king huddled together; the winning method is to put the defender in zugzwang, force the rook to step away from its king, and then win it with a forking check. Everything in this endgame — the defender's third-rank barrier, the attacker's move-passing maneuver — is a fight over that single separation.
Why the Rook Must Stay Glued to Its King
A rook that stands apart from its king in this endgame is already lost, and the geometry is worth seeing once in slow motion: the queen checks, the king must step somewhere, and sooner or later one of the checking squares also attacks the rook. Two pieces, one fork. The entire defense is therefore a huddle — rook within one square of the king, ideally shielded from checks — and the entire attack is a campaign to make the huddle illegal for one move.
Here is the endgame's central position, which I put on the board for every student. White: king c6, queen a5. Black: king b8, rook b7 — the huddle at its tightest. With Black to move, it is zugzwang, and everything loses. The rook's best try is to run far:
1...Rb1 (staying close changes nothing: 1...Rb2 and every other rook move on the file or rank runs into the same machinery, only faster) 2.Qc7+ Ka8 3.Qh7! — not a check but a quiet domination move: the rook is attacked from across the board while the king stays caged. 3...Rb4 4.Qg8+ Ka7 5.Qf7+ Ka8 6.Qf8+ Ka7 7.Qxb4. Look at the finish: every check came on a square that also watched the rook's neighborhood, and the last one, from f8, hit b4 straight down the diagonal — the checks never gave the rook a tempo to run further. Check, check, fork, and the game transposes to king and queen vs king. This seven-move collapse, tablebase-verified as Black's best resistance from the diagram, is what the whole endgame is about: the moment the rook leaves the huddle, the queen's checks hunt down the alignment.
The Zugzwang Handoff
The catch: the diagram above is only deadly with Black to move. With White to move it is a mutual-zugzwang standoff — every immediate try lets the huddle re-form. White would love to pass. Since chess has no pass, White needs a maneuver that returns the same position — or its mirror — with the move handed to Black. Watch the queen walk the triangle:
1.Qd5! threatens mate on a8 and forces 1...Ka8 (the rook cannot leave the seventh rank, and 1...Rb1 walks into the collapse from the first diagram, only faster). 2.Qg8+ Ka7 3.Qd8! Three queen moves, two forced king moves — and White has reached the first diagram's twin: same king on c6, same rook stuck on b7, the queen and the black king each shifted one station, and — the entire point — Black to move. The tablebase scores the two binds identically. The zugzwang has been handed off. The collapse follows on schedule: 3...Rh7 (the least-bad flight) 4.Qb6+ Ka8 5.Qb1! — again the quiet move behind the checks, cutting the rook's file and eyeing the long diagonal — 5...Re7 6.Qf5 Kb8 7.Qf8+ Re8 8.Qb4! and the rook's blockading square is dominated: 8...Kc8 9.Qb7+ Kd8 10.Qc7#. Mate in ten from the handoff diagram, every move tablebase-optimal.
The maneuver generalizes into the endgame's working recipe: fix the huddle with a mate threat, check the king onto a worse square, then retreat the queen along a line the rook cannot contest. Triangulation with checks mixed in — the queen burns a tempo the rook has no way to copy.
The Third-Rank Defense — the Defender's Best Stand
Long before the huddle-in-the-corner stage, the defender has one setup that genuinely earns its keep: king near the edge, rook posted on the third rank in front of the king — for Black, the sixth rank — forming a fence the white king cannot cross. King e8, rook e6 against king e4: as long as the rook patrols that rank, the white king cannot approach without walking into rook checks at full checking distance, and the queen alone cannot build a mating attack. Tablebase verdict on that exact position: still winning for White, but twenty-three moves with best play — the longest stand the defense can make this side of the worst case.
Breaking it requires the same zugzwang tools, applied a phase earlier:
1.Kd5! steps up and asks the fence a question: the queen behind it now eyes a2–g8, and after 1...Kf7 2.Qa2! the pin-threat along that diagonal forces the rook off its rank at once — 2...Rh6, fleeing to maximum distance, because passive squares lose the rook to the approaching king. 3.Qb3 Rh5+ 4.Kd6+! — the point: the king's step is also a discovered check from b3, so Black has no time to re-establish anything. 4...Kg7 5.Qc3+ Kg6 6.Ke6 Rg5 7.Qc2+ Kg7 and the king-hunt continues with the rook drifting ever further from its king — the fork is coming, and the tablebase confirms White is converting steadily. The third-rank fence, once crossed, never re-forms.
Practical Play, for Both Sides
Attacking: centralize the queen, drive the king toward a corner with check-plus-quiet-move rhythms, and hunt zugzwang rather than the rook — the rook falls out of zugzwang, never out of chasing. Watch two safety items: keep a flight square for the defending king at all times (a stalemated king with a loose rook means desperado checks — a rook that can be captured only at the price of stalemate will check forever), and mind the fifty-move count; thirty-five is the worst case with perfect play, and human wandering has burned the budget in famous instances.
Defending: huddle, third rank, and patience. Keep the rook a knight's-move or one square from the king where checks cannot fork; take the third-rank stand when the attacker's king approaches; and remember there is no fortress here — unlike rook and bishop vs rook, which is drawn with correct defense, queen versus rook always loses eventually against perfect play. Your practical target is different: make every step cost moves, and let the fifty-move rule and human error do the rest. Defenders who know just the huddle and the third-rank fence save this endgame constantly at every rating level.
Frequently Asked Questions
Is queen vs rook always a win for the queen?
Yes — with no pawns on the board, every legal position is a forced win, at most thirty-five moves against perfect defense by tablebase count. There is no drawing fortress; the practical draws you see come from the fifty-move rule and attacking mistakes, not from the position.
How does the queen actually win the rook?
By zugzwang, not by chasing. The queen fixes the defending king, hands the move to the defender with a triangulating maneuver, and the rook is forced to step away from its king — after which a short series of checks aligns king and rook and a forking check picks the rook up.
What is the third-rank defense?
The defender posts the rook on the third rank counting from his own side — rank six for Black — with the king behind it, building a fence the attacking king cannot cross without being checked away. It is the toughest known defensive stand, though tablebases confirm it too eventually falls to zugzwang.
Why is this endgame so hard to win in practice?
Because the win runs through quiet moves, not checks. Natural checking sequences let the rook and king re-huddle indefinitely, and the fifty-move rule punishes wandering. The zugzwang handoff — reaching the same bind with the defender to move — is a technique that must be learned deliberately.
What should the defender do against the queen?
Keep the rook next to or one careful square from the king so no check forks them, use the third-rank fence while the attacking king is far, and never send the rook out for checks unless it can return. Every extra move you survive pushes the attacker toward the fifty-move limit.