Despite the obvious efficiencies of many forms of asexual reproduction, sexual reproduction abounds. Asexual species, for the most part, are relatively short-lived offshoots of sexual ancestors. From the nineteenth century, it has been recognized that, since there is no obvious advantage to the individuals involved, the advantages of sexual reproduction must be evolutionary. Furthermore, the advantage must be substantial; for example, producing males entails a two-fold cost, compared to dispensing with them and reproducing by parthenogenetic females. There are a large number of plausible hypotheses. To me the most convincing of these are two. The first hypothesis, and the oldest, is that sexual reproduction offers the opportunity to produce recombinant types that can make the population better able to keep up with changes in the environment. Although the subject of a great deal of work, and despite its great plausibility, the hypothesis has been very difficult to test by critical observations or experiments. Second, species with recombination can bunch harmful mutations together and eliminate several in a single "genetic death." Asexual species, can eliminate them only in the same genotype in which they occurred. If the rate of occurrence of deleterious mutations is one or more per zygote, some mechanism for eliminating them efficiently must exist. A test of this mutation load hypothesis for sexual reproduction, then, is to find whether deleterious mutation rates in general are this high--as Drosophila data argue. Unfortunately, although molecular and evolutionary studies can give information on the total mutation rate, they cannot determine what fraction are deleterious. In addition, there are short discussions of the advantages of diploidy, anisogamy, and separate sexes.