Zugzwang most often occurs in the endgame when the number of pieces, and so the number of possible moves, is reduced, and the exact move chosen is often more critical. A simple example is the following:
Whoever is to move in this position loses the game - they must abandon their own pawn, thus allowing their opponent to capture it and go on to promote their own pawn. Because this is zugzwang no matter who is to move next, this is an example of mutual or reciprocal zugzwang.
The game Fritz Saemisch - Aaron Nimzowitsch, Copenhagen 1923, is sometimes called the "Immortal Zugzwang game" because the final position is widely accepted as being an extremely rare instance of zugzwang occurring in the middlegame. It ended with white resigning in this position:
White has a few pawn moves which do not lose material, but eventually he will have to move one of his pieces. If he plays Rc1 or Rd1 (see algebraic notation) then ...Re2 traps white's Queen; Kh2 fails to ...R5f3, also trapping the queen (white cannot play Bxf3 here because the bishop is pinnedned to the king); g4 runs into ...R5f3 Bxf3 Rh2 mate. Other white moves lose material in more obvious ways. Whether this is true zugzwang is debatable however, because even if white could pass his move he would still lose after ...R5f3 Bxf3 Rxf3, when his queen is again trapped.
See also: null-move heuristic