Under-promotion to a knight?

In over 30 years of playing chess, I had never promoted a pawn to a knight in a serious game.  I finally got to do this recently, and not simply because it was possible, but because it was the best move.

Obvioulsy this type of under-promotion is very rare since a queen is almost always preferable (or occassionally a rook to avoid stalemate).

I'm curious how common under-promotion really is.  Have any of you under-promoted in a serious game, where it was objectively the best move?  And I've mentioned under-promotion to a knight or rook, but is there ever a good reason to choose a bishop instead?

Sorry but the computer says it wasn't the best move. You get mate in 8 with the knight underpromotion, but mate in 5 with 50 Qc4.

This guy does give a plausible reason to choose a bishop underpromotion:

http://www.chess.com/blog/smurfo/the-most-unusual-move-in-chess-or-quothow-i-won-a-beer-in-my-lunch-breakquot

According to this wikipedia article, the frequency of promotions to each piece, based on a ChessBase database, are

However, the article also notes that not all of the underpromotions were necessary or significant.

Thanks notmtwain for the analysis.

I must admit that I suspected that ...c1=N+ was not really "objectively best" (leading to the quickest checkmate).  But it was certainly good enough, as was the less ambitous ambitious ...Qxe3+, Kxe3 c1=Q+.  I had 3 days per move, so I should have tried to find the truly best move, but I gave in to the temptation of the pretty under-promotion (and a bit of laziness).

On the other hand, if it had been an OTB game, I could justify settling for any obviously winning move since my position was so strong.  The 8-move mate, even though longer, is easier to evaluate since I win white's rook instantly.  In other words, practicality is sometimes preferable (safer) than finding the objectively best move, especially if I'm tired and therefor more likely to miscalculate.

Regarding under-promoting to a bishop, I'll read the linked article to condsider the reasoning.  I'm think there must be positions in which promoting to queen or rook would stalemate, but where a knight would give insufficient mating material.

There was a game posted here last week with the same question associated with it ( http://www.chess.com/forum/view/game-analysis/endgame-under-promote-to-simplify ), and I think in that case, the underpromtion was the best move. It allowed the player to fork his opponents king and rook, and the rook (if memory serves) was his opponent's last non-pawn peice.  I think the game would have been longer with a queen while the rook was still in play.

There are 13 serious underpromotions to bishop and 32 to rook at http://timkr.home.xs4all.nl/chess2/minor.htm if I counted correctly.

Thanks ponz111, the game you posted had an interesting opening gambit (Ponziani Opening, Jaenisch Variation), and a very cool finish, a perfect example of appropriate under-promotion to a knight.

Black must not have known any theory, since it has been known for a very long time that 6...d6?? loses on the spot, as you demonstrated.

I would only promote to a Queen or a Rook if I had to, I didn't have the energy to go for an interesting checkmate, or I wanted to learn how to mate with a Queen or Rook more efficiently.

Provided it is safe to do so I would always go for a Knight or Bishop.

I can checkmate with two (opposite coloured) Bishops and with three Knights.

Sadly I can't do the Knight and Bishop Mate yet, but I'd love to learn it. 

http://www.chess.com/forum/view/more-puzzles/plasketts-puzzle

Here is what many players would play from the first position posted here...

There is one rather recent and important top game with an underpromotion to knight, see move 62 here, played by Nakamura when he won against Kramnik in the Chess Olympiad 2012:

http://www.chessgames.com/perl/chessgame?gid=1688934

XDave121X wrote:

http://www.chess.com/forum/view/more-puzzles/plasketts-puzzle

Interesting, the hardest chess puzzle I'll ever see. Even Tal had to think about it long and hard. More difficult than a Zen koan. I want to try. No one post the answer.

I wonder if that Nakamura-Kramnik game isn't hard to beat as "strongest" game ever with an underpromotion to knight that is also the only winning move. There are a bunch of games with underpromotions, but in September 2012, when the game was played, both Nakamura and Kramnik were top five.

Last time I had endgame king+bishop vs king+knight and I had one more pawns. I played e8N+ (attack to king and knight) exchange knight and I won this game (e8Q was bad).

Sometimes if I have 2-3 pawns and my opponent not resign I play endgame K+2B vs K or K+3N vs K.

I just promoted to bishop for the first time in my life in an ongoing turn-based game. It was a curious position where promotion was the best move, yet it didn't matter at all which piece I promoted to. I chose bishop just to cross it off my bucket list. I hope my opponent wasn't too annoyed. :)

this happened to me in a simul game against a GM. promotion to a queen led to a knight fork and an immediate loss but under promotion to a knight held the draw. the trouble was I had it set up for auto promotion to a queen and was playing on the increment only in severe time trouble. I just about sorted it in time and held the draw. havent underpromoted seriously before or since.

I have only underpromoted twice. And that was to a Knight which ended up being the mating piece. Another time to a Rook to avoid stalemate. And this has been over a span of about 18 years.

It was just an (over all badly played) blitz game, and the move is not exactly difficult to spot, but still this is something I've never done before.