Rook endgames with an extra pawn: what decides between the win and the draw

Short answer

In a rook and pawn against rook ending, the result is usually settled before technique enters: if the defending king reaches the zone in front of the pawn, the position is a draw; if it stayed outside, it is a win. This guide measures that zone with endgame tablebases, sets out the Philidor defence and the Lucena position with every line checked square by square, and explains why the defending rook on the eighth rank loses what the same rook on the sixth saves.

The essentials in a few lines

• In a test position with the white king on e4, the pawn on e5 and both rooks free, we checked all 37 legal squares for the black king against the tablebase: nine draw and twenty-eight lose — the difference is having reached the zone in front of the pawn.
• Every drawing square lies two squares or fewer from the queening square; from the c-file leftwards, no square holds the position.
• The same position with the black king on e8 is a draw with the rook on a6 and a loss with the rook on a8: 1.Rh8+ Kd7 2.Rxa8 wins the rook, which is why the Philidor defence needs the sixth rank.
• In the Lucena position the winning technique is the cut along the rank with 2.Rd4 and 3.Rd6+, not walking the king out via the queenside: after 4...Rd2+, only 5.Ke6 keeps the win.
• A draw and a loss are separated by one king square: with the defending king on g7 the position is drawn, with the same king on c7 the tablebase gives forced mate in 29 moves.

The extra pawn is not the deciding variable

This is why material intuition misleads more here than in any other ending. Winning a pawn in the middlegame is an advantage that converts, on average, with work; winning a pawn in a rook ending may be worth nothing at all, and the player who does not know it spends thirty moves chasing a win that does not exist — and, with a short clock, sometimes loses instead.

The practical question, then, is not "how do I convert this pawn". It is "does his king get there?". That question is answered by counting moves before deciding to simplify, and it is the most important decision of the ending: trading pieces with the opponent's king near the pawn usually trades a real advantage for a technical draw.

Every position in this guide was checked against endgame tablebases, which solve exactly all positions of up to seven pieces. Nothing here is an engine evaluation with a margin of error: with five pieces on the board the result is win, draw or loss, and the count to mate is exact. Where there is a number, it was read from that base.

The drawing zone, measured square by square

The experiment has exactly one variable, and that is what makes it useful. It is not a sample of randomly drawn positions, where accidental tactics pollute the statistics: it is the same position repeated 37 times with the black king somewhere else. Everything that separates a draw from a loss is where that king stands.

The shape of the zone is what the practical rule has described for two centuries, with the outline now measured: the squares that hold are the six on the d-, e- and f-files across the last two ranks — d7, d8, e7, e8, f7, f8 — plus the three g-file squares, still within two of e8. From the c-file leftwards there is no drawing square on any rank. Cutting off the king, the classic manual theme, is precisely the manoeuvre that throws the defending king out of that band.

It is worth saying what the experiment does not prove. It holds for a central pawn on the fifth rank, with the rooks on the squares described and White to move; changing the pawn's file or rank changes the zone, and the rook's pawn is a case apart, easier to draw. What generalises is the method: before studying technique, count whether the king arrives.

The Philidor defence: sixth rank, not the eighth

The test position is the same one: white king e4, pawn e5, rook h2; black king e8 and rook a6. The tablebase gives a draw, and the defence is mechanical. While the pawn stays on e5, the black rook holds the sixth rank: the white king cannot enter d6, e6 or f6 without being driven off, and without the king the pawn does not advance on its own.

The moment the plan changes is the one detail that must be memorised. As soon as the pawn plays e5-e6, the sixth rank stops being useful — the pawn now shields the white king from checks along it — and the rook drops at once to the first rank, as far away as possible, to check vertically. That is why the correct defence in the line below plays 1...Ra1 and not something that looks more active.

What the checks from behind do is stop the white king from escorting the pawn. Without the king, the pawn on e7 does not promote: White needs the king on d7 or f7 to guard the queening square, and every attempt to get there runs into a check. Distance matters: the defending rook needs room, which is why it goes to a1 and not to c1.

Same position, rook two ranks higher, and the result flips

The detail is worth more than the example. Plenty of players put the defending rook on the eighth rank by reflex, to keep it away from king and pawn, and the eighth is precisely the rank the defending king is also on. The attacker only has to enter that rank with the rook from the free end — here the h-file — to win the rook with a check.

It is the same idea that makes cutting off the king so strong in this ending: in a five-piece position, almost everything that decides is the geometry of lines. Defending king and rook sharing a rank with the far side open is a lost configuration. On the sixth rank the rook is safe because it shares no line with its own king, and that is the technical reason behind the Philidor defence, more than anything to do with a "third rank".

Note the price of the lapse: two squares separate the position the tablebase scores as a draw from the one it resolves with mate in 10 moves. No engine needs consulting to see this during a game — you only have to ask, before putting the rook anywhere, whether it and the king share a line with the exit open.

When the king is outside the zone: cut, then push

The cut is a one-move decision that is worth the whole game. As long as the white rook holds the d-file, the black king is stuck on the wrong side of the board, and everything else — bringing the king, advancing the pawn, choosing when to trade rooks — happens unopposed. That is why the cut comes first, before any pawn move.

The order matters and is counterintuitive: advancing the pawn early gives the defender exactly what he wants, which is checks and time. In the line below the e5 pawn does not move for nine moves; what moves is the white king, travelling from e4 to f7 while the black rook spends moves trying to annoy. Only once the king has arrived does the pawn go.

The contrast with the previous section is the whole lesson of this guide. Both positions have exactly the same pieces; the difference is that in one the black king is on g7 and in the other on c7. One is an immediate draw, the other is mate in 29. If you ever have to choose between winning a pawn and getting your king in front of the opponent's pawn, take the king.

The Lucena position, and what the check revealed

The idea is economical: the rook goes to the fourth rank while the white king still guards the pawn, then drops to the sixth with check and pushes the black king away. With the black king evicted, the white king steps out via d8, the rook goes to d7 guarding the pawn along the rank, and promotion happens with the black rook forced to trade itself: 4.Kd8 Kf7 5.Rd7 Kf6 6.e8=Q Rxe8+ 7.Kxe8, a whole rook up.

What the check revealed, against what is usually memorised: in this position the bridge manoeuvre with the rook on the fourth rank does not win if it is played after the white king has come out. We ran the sequence 1.Rc4 Re2 2.Kd7 Rd2+ 3.Ke6 Re2+ 4.Kd6 Rd2+ and queried the tablebase on the resulting position: of the six plausible moves, only 5.Ke6 keeps the win. The natural 5.Rd4 allows 5...Rxd4+, drawing, because the white king is on the d-file and the rook is captured with check.

This does not invalidate the bridge as a technique — it exists and it works in other configurations, and the difference lies in which file the attacking king occupies when the rook interposes. What remains is the practical rule: before interposing the rook to block checks, make sure the interposition is not on the same line as your own king. It is the same geometric mistake as in the previous section, made from the other side of the board.

• Before simplifying into rook and pawn against rook, count whether the enemy king reaches the zone in front of the pawn.
• Defending: king in front of the pawn, rook on the rank immediately in front of it, and drop to check only once the pawn advances.
• Never leave defending king and rook on the same rank with the far end open: that is a check, and the rook falls.
• Attacking: cut first, advance the pawn afterwards — an early pawn hands the defender checks and time.
• Before interposing the rook against checks, make sure it does not land on your own king's file.

Frequently asked questions