# How to play this endgame without consulting the tablebases

Recently I got this position in a game I played as Black with White to move.

`````` [FEN "8/8/8/R6p/4KP1P/6P1/pr6/1k6 w - - 0 1"]
``````

Here, White captured with the move 1.Rxa2. It turns out that 1... Kxa2 is winning while 1... Rxa2 loses. Is there any way to see that Kxa2 must be played instead of Rxa2 without turning to the endgame tablebases? How do you judge such endgames?

• 1...Rxa2 is a draw, not a loss. (This doesn’t invalidate the question, of course.) Jul 23, 2020 at 6:49
• There's still a long game to play after either of the moves. That being said, you want your king to reach a square near f8 as soon as possible, and ...Kxa2 puts it one move closer (despite what it may seem in "standard geometry") Jul 23, 2020 at 6:53

In positions where the side with the Rook is trying to hold back the passed pawns it is often critical how quickly the King can scramble back to base, because the Rook will usually be easy. If for example the King must reach g8, then ..Kxa2 leaves the King with one rank less to cross and therefore (other things being equal) will save a tempo. Therefore my instinct would tend toward ..Kxa2 rather than ..Rxa2, even before attempting any analysis. The question whether other things truly are equal is the reason why endings in general need a lot of calculation. In this case they probably are.

However, why did White take the pawn before it promoted? 1.f5 waiting for 1..a1=Q 2. Rxa1 Kxa1. White would then be two tempi up compared to the game. He made Black spend a move on promoting, and also ensured that the Black King would have to cross all eight ranks.