A position with the only legal move resulting in checkmate

Question

Theoretically, is it possible that for some position, the only legal move for a side leads to a checkmate (that is, the opposite king gets checkmated)? If so, has this ever occurred in a real game?

Answer 1

I want to offer a semi-realistic example. I think I have seen something like this in a game by masters in some book, but of course I cannot recall where. But this is something that at least can be realistically missed in calculations from far away.

[FEN "6k1/5ppp/6r1/3b4/4r3/8/1Q5P/1R5K w - - 0 1"]

White just gives a back rank mate, right? :)

Update: actually, I found a real game with a very similar motive, though not quite self-mate in one:

[FEN ""]
[Event "Simul"]
[Site "Copenhagen DEN"]
[Date "1924.03.15"]
[EventDate "?"]
[Round "?"]
[Result "1-0"]
[White "Aron Nimzowitsch"]
[Black "Paulsen"]
[ECO "D32"]
[WhiteElo "?"]
[BlackElo "?"]
[Source "Skjoldager and Nielsen pp.399-400"]
[PlyCount "43"]

1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Nf3 c5 5. e3 Nc6 6. a3 cxd4
7. exd4 Be7 8. c5 O-O 9. Bd3 a6 10. O-O Qc7 11. b4 e5 12. dxe5
Nxe5 13. Nxe5 Qxe5 14. Bb2 Ng4 15. f4 Qd4+ 16. Kh1 Nf2+
17. Rxf2 Qxf2 18. Nxd5 Qxb2 19. Bxh7+ Kh8 20. Qh5 Qxa1+ (20...Bg4!)
21. Bb1+ Kg8 22. Nxe7# 1-0

An old Russian book by Judovich also lists the following position in a game between two amateurs "whitnessed by J. Krejcik in a local club":

 [Fen "8/4N1p1/7k/4B2P/5KP1/3q4/8/7R b - - 0 1"]

The story goes that after 1...g5, both players resigned, because white forgot about the en passant rule, and black only remembered it after they made the move. I think we can safely assume it's made up, though.

Answer 2

Sure it's possible. In fact, it's possible that this kind of "inevitability" happens earlier in the game, with many more moves to go until mate.

Here is an example taken from the Wikipedia on joke chess problems. The composer's full name comes from chess problem databases.

[Title "Vilhelm Röpke, Skakbladet 1942, Mate In 6"]
[FEN "K1k5/P1Pp4/1p1P4/8/p7/P2P4/8/8 w - - 0 1"]

1. d4 b5 2. d5 b4 3. axb4 a3 4. b5 a2 5. b6 a1=Q 6. b7#

Answer 3

[FEN "kq6/8/K7/PQ6/8/8/8/8 b - - 0 1"]

1... Qb7+ (1... Qa7#) 2. Qxb7#

Black could mate with 1...Qa7# or lose with 1...Qb7#, when White's only legal reply is 2. Qxb7#.

Answer 4

Forced mate positions do occur in real games although it is pretty rare: such a mate is only delivered about once in 68,000 games on lichess.org. One example comes from this game with the penultimate position:

[FEN "KQ6/2k5/1qP5/N7/5p1p/8/6PP/8 b - - 0 51"]
[Event "Rated Blitz game"]
[Site "https://lichess.org/1aj0ru8r"]
[Date "2013.01.04"]
[White "captainRobinson"]
[Black "agnefrance"]
[Result "0-1"]

1... Qxb8#

The vast majority of these forced-mate positions in real games involve either a king that is in check with the only available move being to capture the checking piece and in doing so deliver mate or a situation in which the mating side has only pawns and the only legal move is a pawn push that mates the opposing king. Both of these situations almost always occur on the side of the board.

A slightly more interesting example from this game where the king is mated by the only legal move in the middle of the board is:

[FEN "8/1R3p1N/1p1k2p1/1bPP4/2nKP3/r4P2/6PP/8 b - - 0 32"]
[Event "Rated Blitz game"]
[Site "https://lichess.org/a43impad"]
[Date "2013.01.05"]
[White "dnc"]
[Black "Semmel"]
[Result "0-1"]

1... bxc5#

Forced-mate positions occur OTB as well. The most interesting such game I found was played between GM Danilo Milanovic and IM Jasmin Bejtovic at the 41st Bosna Open in Sarajevo 2011. The example is interesting because the last two moves of the game are forced and the second is a discovered checkmate. The full game from here is:

[FEN ""]
[Event "41st IT Bosna 2011"]
[Site "Sarajevo"]
[Date "2011.05.08"]
[Round "5.10"]
[White "Milanovic, Danilo"]
[Black "Bejtovic, Jasmin"]
[Result "1-0"]
[ECO "E70"]
[WhiteElo "2494"]
[BlackElo "2413"]
[PlyCount "139"]
[EventDate "2011.05.03"]

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Bd3 O-O 6. Nge2 Nbd7 7. O-O c5 8. d5
Ne5 9. f4 Nxd3 10. Qxd3 a6 11. a4 Rb8 12. h3 Ne8 13. f5 Nf6 14. Bd2 Nd7 15. Qg3
Ne5 16. b3 Bd7 17. a5 b5 18. axb6 Rxb6 19. Nc1 Qb8 20. Ra3 Qc8 21. Ra2 Qb8 22.
Nd1 Re8 23. Ba5 Rb7 24. Raf2 Bh6 25. Bd2 Bxd2 26. Rxd2 a5 27. Ne3 a4 28. bxa4
Rb2 29. Rdf2 Rxf2 30. Qxf2 Qb4 31. Ne2 Qd2 32. Ng4 Nxg4 33. hxg4 Bxa4 34. e5
Rf8 35. f6 exf6 36. exd6 Re8 37. Ng3 Qd3 38. Kh2 Qxc4 39. Qxf6 Qxg4 40. Qxf7+
Kh8 41. Qf6+ Kg8 42. Qf7+ Kh8 43. Rf4 Qd7 44. Qxd7 Bxd7 45. Rf7 Rd8 46. Ne4 Bf5
47. Nf6 g5 48. d7 Bg6 49. Re7 c4 50. Re8+ Bxe8 51. dxe8=Q+ Rxe8 52. Nxe8 c3 53.
d6 c2 54. d7 c1=Q 55. d8=Q Qf4+ 56. Kg1 Qc1+ 57. Kf2 Qf4+ 58. Ke2 Qe4+ 59. Kd1
Qg4+ 60. Kc1 Qc4+ 61. Kd2 Qb4+ 62. Kd3 Qb3+ 63. Ke4 Qc2+ 64. Ke5 Qc5+ 65. Ke6
Qe3+ 66. Kf7 Qb3+ 67. Kf8 Qb4+ 68. Nd6 Qf4+ 69. Nf7+ Qxf7+ 70. Kxf7# 1-0

These games were found using CQL 6 with the query:

cql()
mate
parent : move count legal == 1

Answer 5

I was thinking about position where your king is not checked, and this seems to work (of course white to play):

[FEN "b5k1/RN3ppp/P7/8/8/8/8/5qbK w - - 0 1"]