The Tarrasch Rule and Rook Endgames
Contents
Rook endgames are the most common of all: roughly one out of every two endgames that survive into a reduced-material phase features a rook on each side. And yet they are also the most poorly played, even at the highest level. The reason is simple: the rook is a long-range, dynamic piece whose value depends less on where it stands than on what it targets. A passive rook is worth a minor piece; an active rook is worth almost a win.
The whole secret rests on an intuition that Siegbert Tarrasch boiled down to a now-proverbial phrase: the rook belongs behind the passed pawn. Behind your own to push it, behind your opponent’s to restrain it. Understand this rule, add the two key positions — Lucena to win, Philidor to draw — and you will already steer your endgames better than the vast majority of players at your level.
This article gives you that bedrock, with diagrams to anchor it. Not a list of variations to memorize, but five guiding ideas you will recognize on the board, game after game — together with their exceptions, because in rook endgames nothing is ever quite absolute.
The Tarrasch Rule: the rook behind the passed pawn
Picture a passed pawn pushing forward. If your rook stands behind it, the rook follows square by square: it keeps its full range, and the pawn gains ground without ever losing its protection. If instead your rook is in front of the pawn, it is forced to retreat with every advance — it loses range, and it blocks its own pawn’s path. That is the whole point of the Tarrasch rule: the rook belongs behind the passed pawn.
The corollary matters just as much on defence. Place your rook behind the opponent’s passed pawn, and the further that pawn advances, the more exposed it becomes: your rook controls the whole file and rains checks on the king if it tries to escort the pawn. A rook restraining a pawn from behind, keeping its activity, is worth far more than a rook stopping it head-on, condemned to passivity.
Keep the practical formula in mind: behind mine to push it, behind yours to restrain it. That single sentence settles most rook-placement decisions in the endgame.
A strong rule, not an absolute law
Tarrasch himself qualified his maxim: it is true in general, not always. Two classic situations show why. With a rook’s pawn (the a- or h-file), a rook behind the pawn can lack the room to manoeuvre, and the frontal defence — rook in front of the pawn — sometimes proves more tenacious.
Likewise, against a pawn still far from promotion, keeping your rook active on the flank to deliver side checks can be better than gluing it behind the pawn. The Tarrasch rule stays your default reflex, right about 90% of the time; but in rook endgames, always check that no exception applies before following it blindly.
The outside passed pawn, the decisive weapon
Why do we care so much about these passed pawns? Because they pull the enemy king away. A passed pawn far from the main body of forces forces the defending king to chase it down, away from the real battlefield. Meanwhile your own king feasts on the pawns left unguarded on the other wing.
This is exactly the mechanism at work in the position below. The Tarrasch rule and the outside passed pawn work hand in hand: the rook pushes the pawn, the pawn draws the king, and the endgame tips over.
The Lucena position: the method to win
Here is the position every player should know cold. You have a passed pawn that has reached the 7th rank, your king stands in front of it but is boxed in on the promotion square, and the enemy rook keeps harassing you with checks. How do you promote without losing the pawn? The answer is called Lucena, and the technique has a vivid name: "building a bridge."
The idea is to place your rook on the fourth rank from your side, then walk your king out from in front of its pawn. When the enemy rook comes to check, your rook will interpose — the "bridge" — and shut the checks down for good. The pawn then queens.
Lucena is THE reference winning position of rook-and-pawn endgames. If you were to memorize a single attacking technique, this would be the one.
The rook-pawn exception, and the three-file rule
A warning: Lucena does not win with a rook’s pawn, on the a- or h-file. The attacking king, posted in front of its rook’s pawn, simply has no room to sidestep and build the bridge — the edge of the board boxes it in. Many a/h-pawn-versus-rook endgames are therefore drawn despite the extra pawn, where the same setup on a central file would win.
The other practical key to the win is to cut the enemy king off from the promotion square. The textbook rule of thumb: cut it off by at least three files. With that much distance, the defending king never arrives in time, you calmly reach Lucena, and the bridge does the rest.
The active king, precondition of every promotion
Even before the bridge, a more fundamental truth: a pawn almost never queens alone. It needs its king’s escort. In the simplest endgames, without rooks, a king marching in front of its pawn already guarantees the win — it clears the promotion square and keeps the enemy at bay.
Do not confuse this active king with the opposition, a neighbouring but distinct idea: here, what matters is that the king leads its pawn and opens the road, not a face-to-face of squares.
The Philidor defence: the method to draw
On defence, the mirror position is called Philidor, and it is every bit as essential. You have a rook against rook-and-pawn, your king stands in front of the enemy pawn (on the promotion square or just before it), and you must hold the draw. The technique fits into a disarmingly simple instruction.
As long as the enemy pawn has not crossed your third rank, keep your rook on that third rank. It denies the enemy king any foothold there: the pawn’s advance cannot be supported. The moment the pawn finally steps onto your third rank, your rook drops at once to the back, to your first rank, and bombards the enemy king with checks from behind, which it can no longer escape.
This is the Philidor defence. Executed properly, it holds any rook-and-pawn-versus-rook endgame in which your king has reached the promotion square in time.
Short side, long side: where to put the defending king
When the third-rank defence fails because the pawn is too far advanced, a second resource exists: defending with side checks. But it requires choosing the right side for your king, and this is where the "short side / long side" nuance comes in.
The principle: the defending king takes refuge on the short side of the pawn — the wing closer to the edge — so that its own rook gets as much room as possible on the other side, the long side. With that space, the rook delivers repeated side checks the attacking king cannot escape. Stuck on the wrong side, the same rook would be too close to the enemy king and get chased away. Remember the formula: the king goes to the edge, the rook keeps the open space.
Lucena or Philidor: the king’s race decides
How do you tell whether a rook-and-pawn-versus-rook endgame is won or drawn? It all comes down to one question: has the defender’s king reached the promotion square (or the pawn’s file) in time? If it is there, it is Philidor and a draw. If it was driven off or never made it, the attacker reaches Lucena and wins.
Your practical compass during the game: if you are defending, rush the king in front of the enemy pawn as early as possible. If you are attacking, cut the enemy king off from its promotion square before anything else — by at least three files, if you can. The line between win and draw is often a single tempo.
Counting races: the rule of the square
Rook endgames often boil down to a race: will my pawn queen before his king catches it, and vice versa? To settle these races without calculating move by move, use the rule of the square.
Mentally draw the square whose side runs from the pawn to its promotion square. If the enemy king is inside that square — or can step into it on the move — it catches the pawn. If it is outside the square and it is the pawn’s turn, the pawn runs and promotes. One caveat only: if the pawn is still on its starting square, it can leap two squares, which enlarges its square by one rank. Always count that in.
The traps to avoid in practice
Knowing Lucena and Philidor is not enough if you stumble over the details. Two pitfalls recur constantly at your level, and both stem from inattention to the enemy king and rook.
The first: the passive rook. Many players plant their rook in front of the enemy pawn to stop it, condemning it to do nothing. Remember Tarrasch: behind the pawn, never in front — bar the rook-pawn exception we saw. An active rook restraining from behind is worth two rooks glued to defence.
The second, more brutal still: the back-rank mate. In the rush to push pawns and chase promotion, you forget that your own king is walled in on its last rank. A single enemy rook is enough.
In your games
To anchor these ideas in YOUR games rather than in the abstract, follow this path on ChessPivot. First, open your statistics page: the weaknesses section tells you, as a percentage, whether you let wins slip ("missed win," "missed material gain") or mishandle your pawn races. That is your starting diagnosis.
Next, take one of your rook endgames on the analysis page. At the inflection points — the moves where the evaluation tips — the coaching explains in plain terms what changed: a rook misplaced in front of the pawn, a king arriving too late in front of the enemy pawn, an enemy king poorly cut off from its promotion square, a forgotten back-rank mate. Cross-check what you read against the Tarrasch rule and the Lucena/Philidor diagnosis from this article.
Finally, for any vocabulary that escapes you (bridge, opposition, outside passed pawn, short side), the glossary gives you the precise definition, and the blog offers further endgame articles to take it deeper.