Free online Chess server. Play Chess now in a clean interface. No registration, no ads, no plugin required. Play Chess with the computer, friends or random opponents.
Sign in
Reconnecting
  1. Forum
  2. Game analysis
  3. Rook vs knight: now what?

I am interested in this position:

lichess.org/4Rgityc1/black#120

I understand that in general this is a theoretical draw. However, I also know that proving it may be tough for the side with the knight. Any tips on how black can test white most effectively here?

Thanks!

I tried it once in 2018 in a tournament game. It turned out to be absolutely hopeless. You even can't force the King to the edge (which would be still a draw except for some rare cases)

It is easy as cherry-pie to preserve the draw.

PS: actually separating K and N is not a particular good idea for the defender, this might be quickly lost.

It is a clear draw. Accept it.

54...Rc4 and Ra4+ on Ka5 that looks won. You will win the g5 pawn anyway (or the other two), so you can first prevent his counterplay.

The only way to win a Rook versus Knight ending (playing the side with the Rook) is if the Knight is driven too far away in order to intervene to what will essentially be a Rook versus lone King endgame OR the Knight gets trapped on the edge of the board and then captured.

In both cases, the opponent's King and Knight must be separated, while your own King will need to be strategically placed near the piece getting mated/captured... An opponent knowledgeable of the endgame will be able to prevent this 100% of the time, given enough time on the clock. An exception occurs when the Knight is in one of the 4 corners (or 1 square diagonally from each of those), where its mobility is greatly reduced. Then, not even a nearby King will be able to help it from being captured.

I've been working on an endgame study which covers some of this, see the analysis in the chapter below (unfortunately can not link to positions in variations). Hopefully it can be of use to others as well.

lichess.org/study/FbrvzCip/plCMcBM9

This topic has been archived and can no longer be replied to.