S U M M A R YSeveral previous studies have shown that the proportion of clover in sheep diets was greater than the proportion of clover in the mixed species swards being grazed. Various hypotheses have been put forward to explain this observation. One widely held hypothesis is that sheep prefer (actively select) clover to grass. This hypothesis was tested by offering 12 sheep, six that had been recently grazing perennial ryegrass and six that had recently been grazing white clover, the choice between foraging on grass or clover provided as turves. It was found that, rather than prefer clover, the sheep preferred the opposite species to the one they had previously grazed. From these observations, alternative hypotheses are proposed to explain the disparity in composition of the diet relative to the swards observed in previous studies, and the evidence for a constant preference for clover and its basis in grazing behaviour is reconsidered.