Best UnderPromotion Endgame Compilation

Many of those moves aren't forced though.

Nice one, it seems pretty hard for black to hold with queen due to White's extra pawns, especially the one defending e4.

I'm not entirely sure what you mean, but all of white's moves are the best (and only) move.

The king has other squares at multiple points, and there's even a point where black can just attempt to play with a queen vs 2-3 minor piece endgame. Maybe I'm missing something, can you post the variations?

A simple rook vs 2 pawn endgame with a simple underpromotion tactic at the end. After 10...Ra1, if white promotes either pawn to a queen, black mates white next move. If white plays Kf8, black mates on a8, if white plays kd8, black takes the f7 pawn and then gets the g7 pawn next move, and wins. But, after f8N+, deflecting the king off e6, white will easily win the endgame. The move Kc6 is very subtle, it is to ultimately give the king an escape to the 5th rank eventually if black continue to hunt the pawns with his rook, (if the white king retreated to e7 black would be able to capture g7 with a pin and sac for the f7 pawn next move, draw). Simple underpromotion but may be missed in a blitz game, so take note of this one!

It's incorrect. If white only wanted a draw he could have queened. So obviously white must win but instead he loses a bishop.

I suppose the diagram is wrong. It's not too hard to change it into a win but the challenge is to do it in an interesting way!

Interestingly I made a similar study a few years ago - expecting an earlier one would exist. This one however is dualed in a pretty fundamental way. 3.Ke6 (instead of 3.g6) wins not only faster but avoids having to find the tricky white king maneuver and the underpromotion.

Here is a double underpromotion:

1.g8=N+ Kh7 2.f8=N+ Kh8 3.N4g6#

My bad, the bishop should be on e3, then it is a win as black has no satisfactory answer to Be4+ Kh8 Bd4#

That makes sense. And the complete solution is better as well!

Not 1, not 2, but 5 bishop promotions in 1 problem! Queen promotions lead to a stalemate after the rook captures h5 and the queen is forced to recapture back. The dark squared bishop simultaneously guards a1 from the rook trying to get there and mate on f1, while also making it harder for black to play rh8 at some points. Bd4 is a waiting move which also hits f2, and thereby prevents kh4 escaping. After the 5 bishops white can promote to a couple queens and easily wins!

5 Bishop promotions, impressive! And the position is legal too!

4 different promotions.

Perfect Puzzle. Here's another simple one yet funny. The position is also just barely legal as both sides have the maximum number of pieces possible with the number of pawns and material count:

I feel like there has to be some other way for white to win here, how could this be the only solution?

This beautiful AUW (AllUmWandlung = 4 different promotions) endgame study is by Harold Lommer probably published in 1933. It's not in his Arves biography which suggests it contains an error. However I can't find one, even with engine help. Lommer made several studies with multiple pawn promotions so it is in style!

Btw, due to the quality of the underpromotion puzzles in this topic I am surprised we haven't seen the Babson task here yet. Somebody is going to post that, I presume???

Due to the black discovered checking threats by the giant Rg6 white is forced to hand out checking moves himself. It appears that the knight promotions are the only way to achieve that.

I found it in "Lehr- und Handbuch der Endspiele vol. 3 by Andre Cheron (1958)". My engine couldn't find errors. But Cheron gives a lot of annotations. Maybe there is a glitch in the sidelines. Didn't check them out.

Another one. But it is even more difficult, there are 6 pages of analysis.

It's impressive what people analyzed back then, without engines.