You improve at endgames by learning a small number of exact positions in the right order, not by reading general advice. The order that works is: king and pawn first (the square of the pawn, the opposition, key squares), then rook endgames (Philidor to draw, Lucena to win), then the basic mates, then minor-piece basics and the wrong-coloured-bishop draw. Everything else in the endgame is built out of those pieces.

Below is that curriculum, with a board for every named position so you can see the geometry instead of imagining it. Five general rules run underneath all of it, so we start there — but the rules only pay off once you know the exact positions they are pointing at.

The five rules that run through every endgame

These are the habits. They will not win a technical position on their own, but breaking them will lose one.

1. Activate your king. In the opening and middlegame the king hides. In the endgame it is a fighting piece, roughly as strong as a minor piece. March it toward the centre or toward the pawns as soon as the queens come off. A passive king in a simplified position is usually a lost game — see king activity in the endgame for the standard routes.

2. Create a passed pawn. A pawn with no enemy pawn in front of it or on either adjacent file is the single most valuable asset in an endgame, because the threat of promotion ties down enemy pieces even if the pawn never queens. Push your majority: three pawns against two on the queenside should become a passed pawn. Our guide to passed pawns covers how to make and escort them.

3. Rooks belong behind passed pawns. The Tarrasch rule, and it works for both sides. Behind your own pawn, the rook's scope grows as the pawn advances. Behind the enemy pawn, the pawn walks toward your rook and gets easier to stop. A rook in front of a passed pawn is a rook doing nothing.

4. Cut the enemy king off. A rook on a file or rank between the enemy king and the action turns a fair fight into a two-against-one. Most rook endgame wins are really wins by cut-off: your king and pawn work while theirs watches from behind a wall.

5. Do not hurry. Before you commit to a pawn push that can never be taken back, improve every piece to its best square first. Pawn moves are the only irreversible moves in chess. If you cannot see a clear improvement, make a waiting move and let your opponent commit first — that is the whole idea behind zugzwang and triangulation.

Stage 1: King and pawn endgames

Start here, always. Pure pawn endings are the only endings that are fully calculable, and every other endgame eventually simplifies into one. Three ideas cover most of it.

The square of the pawn

Can a king catch a running pawn? Draw an imaginary square whose side is the distance from the pawn to its promotion square, extending toward the defending king. If the defending king is inside that square — or can step into it on its move — it catches the pawn. If not, the pawn promotes.

The square of the h5-pawn runs h5-h8-e8-e5. Black to move is on d5, one file outside it, and loses the race: 1...Ke6 2.h6 Kf7 3.h7 and the pawn queens. Play this position →

Two details decide most practical cases. A pawn still on its starting rank counts from the third rank, not the second, because it can jump two squares. And the square only works if nothing blocks the king's diagonal run — your own pawns getting in the way is the classic way to miscount. The full method is in the rule of the square.

The opposition

When the two kings stand on the same file, rank or diagonal with an odd number of empty squares between them, the player who does not have to move holds the opposition. The player to move must step aside and concede ground, because kings can never touch.

The fundamental pawn ending. White to move only draws; Black to move loses. The result flips purely on whose turn it is. Play this position →

Work through both sides of that diagram, because it is the position everything else reduces to:

White to move (draw):
1.Kd5 Kd7 2.e5 Ke7 3.e6 Ke8 4.Kd6 Kd8 5.e7+ Ke8 6.Ke6 stalemate

Black to move (White wins):
1...Kd7 2.Kf6 Kd6 3.e5+ Kd5 4.Kf5 Kd4 5.e6 Kd5 6.e7 Kd6 7.e8=Q

Distant opposition (three or five empty squares) and diagonal opposition work identically — the side not to move wins the standoff. Opposition in chess drills the counting.

Key squares

Opposition is a tool; key squares are the target. A key square is a square that wins for the attacker regardless of whose move it is. For a non-rook pawn on the second, third, fourth or fifth rank, the key squares are the three squares two ranks in front of the pawn. Get your king onto one of them and you no longer need to calculate.

With the pawn on e4, the key squares are d6, e6 and f6. White's king already stands on one, so White wins whoever is to move. Play this position →

Once the pawn reaches the fifth rank, the three squares directly in front of it join the list. This is why the practical instruction in pawn endings is always king first, pawn second: pushing the pawn ahead of the king throws away the very squares the king needed.

Rook pawns are the exception

An a-pawn or h-pawn has no useful key squares. If the defending king reaches the promotion corner, the position is drawn no matter how much extra king activity the attacker has — there is no second flank for the attacking king to come around from, so every attempt ends in stalemate. Remember this before you trade into a "winning" rook-pawn ending.

Stage 2: Rook endgames

Rook and pawn endings are the most common piece endings in real games, which makes them the highest-return study of any stage. Two positions carry most of the load: Philidor to draw a pawn down, Lucena to win a pawn up.

Philidor: the defensive method

Analysed by François-André Danican Philidor in the eighteenth century, this is the drawing setup you aim for whenever you are a pawn down in a rook ending. Two conditions: your king sits in front of the enemy pawn, and your rook sits on your third rank — the sixth rank if you are Black.

Philidor's drawing position. Black's rook on the sixth rank denies the white king d6, e6 and f6, so White cannot make progress without pushing the pawn. Play this position →

The rook on the sixth rank is a fence: the white king cannot cross it, and a pawn cannot advance without king support. White's only try is to push the pawn to e6 — and the instant the pawn arrives on the sixth rank, it blocks its own king's shelter. Black then drops the rook to the first rank and checks from maximum distance:

1.e6 Ra1! 2.Kd6 Rd1+ 3.Kc7 Rc1+ 4.Kd7 Rd1+ ...

White's king can never escape the checks, because the further it runs from the rook the safer the rook is. Do not check early: leave the rook on the sixth rank until the pawn commits. The complete treatment is in the Philidor position.

Lucena: the winning method

The Lucena is the flip side — the position you steer toward when you are the one with the extra pawn. It is named after the fifteenth-century Spanish writer Luis Ramírez de Lucena, though the position does not actually appear in his surviving book. Its ingredients: the pawn on the seventh rank, your king on the promotion square in front of it, and the enemy king cut off by your rook.

The Lucena. White's rook on e1 cuts the black king off from the pawn; White's problem is that his own king blocks the promotion square. Play this position →

White cannot simply step aside, because the black rook checks from a distance forever. The solution is building a bridge: lift the rook to the fourth rank so it can later interpose, walk the king down toward it, and use the rook as a shield.

1.Re4! Ra1 2.Kd7 Rd1+ 3.Kc6 Rc1+ 4.Kd6 Rd1+ 5.Kc5 Rc1+ 6.Rc4!

That last move ends the game. If Black captures with 6...Rxc4+, then 7.Kxc4 and the pawn promotes with the black king hopelessly far away. If Black declines, the check is blocked and c8=Q follows next move. The fourth rank matters: build the bridge on the third rank and after the rook trade your king is too close to the enemy king to escort the pawn home. Full walkthrough in the Lucena position.

The two rules that decide the rest

Almost every other rook ending is judged by two questions. First, is the enemy king cut off, and by how many files? A king cut off by three or more files from a passed pawn is usually lost. Second, is your rook active? An active rook behind a passed pawn, or checking from behind, is worth more than a pawn in these endings — passivity is what loses them. Rook and pawn endgames covers the practical middle ground between Philidor and Lucena.

Stage 3: The basic mates

You need these to be automatic, because you will reach them with seconds on the clock and a fifty-move counter running.

King and rook against king. Use the box: cut the enemy king onto a shrinking rectangle with the rook, then shrink it by bringing your king up and stepping the rook forward whenever the kings face each other. The finish always looks like this:

The end of the box method. The kings are in direct opposition, so Ra8 is mate — the rook covers the whole eighth rank and White's king covers d7, e7 and f7. Play this position →

Note that the mate needs the opposition — the same idea from pawn endings, reused. Step-by-step in king and rook checkmate.

King and queen against king. Drive the lone king to the edge using the knight-move technique — keep the queen a knight's move away from the enemy king — then bring your own king up and mate. The one thing to watch is stalemate: never take away every square without giving check. See king and queen checkmate.

The mates you can skip. King and two knights cannot force mate against a lone king — mate exists but cannot be forced, so if that is all you have, the game is drawn. King, bishop and knight can force mate, but it takes over thirty moves of precise technique and almost never occurs. Learn it last, if ever.

Stage 4: Minor-piece basics

Now that pawn and rook endings are solid, minor pieces become readable. The whole subject reduces to one comparison: a bishop is a long-range piece that only touches half the board, and a knight is a short-range piece that touches everything but travels slowly.

One bishop, two wings. From d1 the bishop watches a4 and h5 at the same time — a knight on d1 could not even reach either pawn's path in time. Play this position →

That diagram is the entire argument for the bishop. Pawns on both wings favour the bishop, because it defends and attacks on opposite flanks from one square. The knight has to physically travel, and a knight needs three moves to cross the board that a bishop crosses in one.

The reverse case is just as important. A closed position with pawns on one wing favours the knight, because the knight can hop over fixed pawn chains and can attack a pawn sitting on a square the bishop can never touch. A bishop whose own pawns are fixed on its colour is a "bad bishop" — nominally a piece, practically a spectator. Bishop versus knight endgames works through the deciding features, and minor piece endgames covers the mixed cases.

Two other minor-piece facts worth knowing at this stage: opposite-coloured bishops are strongly drawish in pure endings — even two extra pawns often fail to win, because the defending bishop holds squares the attacker can never contest — and knight endgames follow king-and-pawn rules more closely than any other piece ending, which is another reason to learn stage 1 properly.

Stage 5: The wrong-coloured bishop

This is the single most valuable draw in the endgame, and the reason so many "clearly winning" positions end in half a point.

If you have a bishop plus a rook pawn, and your bishop does not control the pawn's promotion square, the defending king draws simply by sitting in the corner. There is no way to evict it, because every attempt is stalemate.

A light-squared bishop and an h-pawn. The promotion square h8 is dark, so the bishop can never control it — Black's king walks to the corner and the game is drawn. Play this position →
1...Kh7 2.Kf7 Kh8 3.Kg6 Kg8 4.h7+ Kh8 5.Kh6 stalemate

The practical consequences are what matter. As the attacker, never trade into this: check the colour of the promotion square before you exchange your last non-rook pawn. As the defender, this is a lifeline — if you are losing badly with a bishop-and-rook-pawn ending on the horizon, head for the corner and trade off everything else. The same corner logic underpins many fortress positions and the stalemate tricks that save lost games. Deeper coverage in the wrong-colour bishop.

How to actually practise this

Reading about a position is not learning it. The method that works is short and repetitive:

  • Play the position, do not admire it. Set up each diagram above against the engine and play it out from both sides until the result is never in doubt. Ten minutes on the fundamental pawn ending is worth more than an hour of reading.
  • Learn one position per session. The list here is roughly eight positions. At one a day you own the entire practical endgame in under two weeks.
  • Check your endgames after every game. When you review, do not stop at the opening. Find the moment the queens came off and ask what stage-1 or stage-2 position the game was heading toward. How to analyse your games has a workflow.
  • Simplify deliberately. Once you trust your technique, trading into a known ending becomes a weapon rather than a risk. That is the real payoff, and it is what converting an advantage is about.

If you are still shaky on notation while working through the lines above, how to read chess notation will take five minutes to fix that.

Practise these positions on the free board

Every diagram on this page has a Play this position link that drops the position straight into the free board at LocalChess. Start with the fundamental pawn ending, then Philidor, then Lucena — those three cover most of what actually decides club games. Play each one from both sides against the engine, and keep replaying until you can produce the drawing or winning method without thinking about it. That is what "good endgame technique" means in practice: not deeper calculation, but recognising the position you are heading for several moves before you get there.