Mate in 34

This was quite an interesting position. At this point, my opponent (playing white) said to me "I'm not sure what's going to happen - either it's a draw or I win." What happened was actually the option he hadn't considered.

On the surface, it does look like white has the advantage. But no - according to Bringer, this is mate in 34 for black.

Updated: this was black to move. Actual game (which had some inaccuracies) follows:

Your diagram says "White to move."  If it's really white to move, I don't believe you. If it is black's move, it might be fun to see.

Would you mind posting your analysis?

Sorry, you were right - it was black to move. I put in the actual game, which wasn't the most accurate, but the Bringer program also played this position to a win for black.

It looks pretty good but I wouldn't trust any computer to analyze an endgame of this complexity.

I would want to try 41 Na1 and try to hold the black king back with my king. Getting a passer wouldn't be easy with the knight able to check the king whenever it went to d4.

It doesn't look like white has the advantage on the surface, really.

That passed pawn is bad news for black.

Hmm, 41. Na1 seems promising, but then the white king is split between trying to protect the knight and defend against the pawn onslaught. Or am I missing something?

The interesting thing about this game happens around turn 30. At that point, I was hoping for a draw, but it look like it was a winning position as far as I can tell. Maybe I can credit some intuition here, but think I just got lucky.

Good point. That would be such a painful way to lose, smothered by your own pawn one square from promotion. I tried it the other way too, where black goes after the pawns after na1 and it looked like black still wins, but I'm not trusting my analysis.

I meant that after 41.. Kd4, 42 Nb3+ . If ..Kc4, then N back to a1. If Ke5 or elsewhere, then Kd2.  

I meant to use the Nb3+ as a way to drive the black king back, since if you ignore the knight and a pawn, the king and four pawns versus king and three pawns wouldn't win unless the side with four pawns had a good king position. 

This type of thing has to be covered in one of the endgame books.

If black tries to push his pawns, at some point, he would end up with a king trying to shepherd a single pawn trying to queen.  If we continue to ignore the knight and a pawn, unless he has the opposition, this would normally end up a draw. However, once the position is reduced to a single pawn, perhaps black can abandon it at the right time and go after the knight.

If you go to the Nalimov table bases, which cover endgames with up to 6 pieces with 100% certainty, and reduce the pawn advantage to one remaining black pawn on d4, the table bases say it is a draw, with either side to move.