Sex is a general feature of the life cycle of eukaryotes. It is not universal, however, as many organisms seem to lack sex entirely. The widespread occurrence of sex is puzzling, both because meiotic recombination can disrupt co-adapted combinations of genes, and because it halves the potential rate of reproduction in organisms with strongly differentiated male and female gametes. Most attempts to explain the maintenance of sexuality invoke differences between parents and sexual offspring. These differences may be advantageous in novel or changing environments if new gene combinations are favoured from time to time. Sex would then serve to concentrate beneficial mutations that have arisen independently into the same line of descent. But in a stable environment sex might serve to concentrate deleterious mutations, so that they will be more effectively purged from the population by selection. We have studied the effect of sex on mean fitness in experimental populations of the budding yeast Saccharomyces cerevisiae. Our results show that sex increases mean fitness in an environment to which the populations were well adapted, but not in an environment to which new adaptation occurred, supporting the hypothesis that the advantage of sexuality lay in the removal of deleterious mutations.