Yes it has happened with GM Nakamura on the losing end of it in St. Louis against GM Ponomariov in May 2011. Both were over 2700 FIDE at the time. LINK
N+B+K v. K - has it ever happened?
Yes, it has happened. I believe there is actually an extension of the 50-move rule in that case. I know someone who drew (I think an IM?), because the IM (?) was up a knight and bishop, but my friend managed to trade off all the pawns, and the IM (?) didn't remember how to do the checkmate. But I know it's really rare.

Wow, thanks KingsEye, I didn´t know that there´d been such a prominent loser!
9thEagle - I can´t imagine there being an extension of the 50-move rule in this case. I believe it´s possible with the worst-case starting position in 33 moves or so, and if I practise it against the computer, mess up and overstep the 50 it´s a draw.

Actually it does not take many moves to check mate with King Knight and Bishop. But you must know the theory behind it and force the opposing King into the corner of the Bishop's color. I get this problem on chessmagnet school and some times I get it right other times I let the king get away. But I feel will never encounter this end game, but non the less its good to know and be able to pull it off and improve your thinking skills.

Depends what you mean by "not many moves"! Ponomariov would have taken 19 if Nakamura hadn´t resigned, but the opposing king was already at the edge of the board, and Ponomariov´s pieces were already perfectly placed after 80.Kxb4 to reach the classical starting position after black´s 83rd move. Had the black king been in the centre of the board, it would have taken a while longer (PGN download at the bottom of the linked page in KingsEye´s post #2, many thanks!).
Here´s the game:

I was on the short end of this recently and made it past fifty moves and claimed a draw. Key is avoid that corner like the plague and its VERY hard to force the king in unless you really know the move pattern.

Just had a look in the Icofy base.
I found 912 games with this material, 208 times it turned into a draw.
A lot of people get this wrong, lol

I've won with it a few times.
I mean I had to sack a few pieces to be left with B&K, but still...
Yeah, that's fun to do :)

Yes, it has happened. I believe there is actually an extension of the 50-move rule in that case.
NO there isn't aside from the fact this mate takes 34 moves max from the winning side's "worst starting position" FIDE abolished ALL exceptions to the 50 move rule in 1992. As the old perfessor Casey Stengle used to say "You can look it up" (but I already did)
You can also look up how to do this mate it's not that hard once you know the technique and remember 1 or 2 "trick" (far from obvious) moves - and you'll amaze your friends, family and chess chums by being able to solve a "Mate in 34." Honest - no kidding - OK I'm lying, but anyways...
http://blog.chess.com/NimzoRoy/mate-with-kbn-vs-k---together-again-for-the-first-time

"You will end up scoring a lot more points if you devote the same time and effort into increasing your proficiency at Rook and pawn endings, which occur in nearly half of all endgames."
It only took me 15 minutes, hardly a big waste.

http://www.chessgames.com/perl/chessgame?gid=1653900
Aronian resigned against Carlsen when he saw the KBN vs K ending coming.
I have only had this endgame once in an online game, but it's very easy once you know how to do it.

I've won with it a few times.
I mean I had to sack a few pieces to be left with B&N, but still...
I´ve not had the pleasure yet, although I had a 2-bishop one a couple of days ago after unobtrusively letting him eat my remaining pawns while I got rid of all of his. B+N would be more fun though.
Thanks for all the feedback folks, especially @rooperi for Epishin´s Immortal (cockup ...)!

A few of them, some smoother than others (all from the same day actually):
http://www.chess.com/livechess/game?id=391947403
http://www.chess.com/livechess/game?id=391907512
http://www.chess.com/livechess/game?id=391896316

I remember the game with Aronian and Carlsen...the commentators wanted to see Carlsen attempt the mate...we were all disappointed to see Aronian resign.

"You will end up scoring a lot more points if you devote the same time and effort into increasing your proficiency at Rook and pawn endings, which occur in nearly half of all endgames."
It only took me 15 minutes, hardly a big waste.
Well, it probably took me a little longer than 15 minutes, but I agree, it's not a waste of time. There is substantial collateral benefit, in many middle- and endgames where R+B have to cooperate. And it is a neat party trick.

I think it's extremely important to learn how B + N coordinate so I believe everybody should really spend as much time on it as necessary to master it 100%.
It really helps your general chess imo.
I even tried to learn the much more complicated Q vs R ending, but eventually I gave up because there's not really recognizable patterns and I felt I was wasting too much energy.

Yeah, I've looked a lot at rook endings, but never run into the book positions in my games, but the concepts are still helpful to know. Any idea how I could have played this better? http://tinyurl.com/ao2cbm2
Andyeah Q vs R is a really tough endgame. I even find Q vs N harder than BN. BN I found pretty simple, from the corner it's a forced series of moves and getting to the corner is a lot more tricky but not massively complicated.
I think I read once that this ending occurred between two masters (IMs or GMs, I can´t remember), and that the guy with the two minor pieces messed it up.
Are there any recorded tournament/championship games at a very high level (say 2500+) with this ending, won or drawn?