The dimorphisms in morphology and behaviour of male fig wasps are among the most extreme in the animal kingdom, and offer excellent opportunities to test the predictions of certain sexual selection models. Winged males resemble their conspecific females closely, but wingless males are so divergent in form that they have repeatedly been classified into different taxa. Wingless males mate within their natal fig fruits, whereas winged males disperse to mate. Individual species may have winged males, wingless males or both morphs. A key hypothesis proposes that sexual selection on male mating opportunities favours winged males in species with small broods and wingless males in species with large broods. Using data from 114 species in 33 genera, we show that both simple and formal comparative analyses support the correlated evolution of large brood size and male winglessness. Theoretical models further predict that, in male dimorphic species, the proportion of winged males should equal (in cases without local mate competition) or exceed (in cases with local mate competition) the proportion of females developing in fig fruits without wingless males. These predictions are met by eight out of nine male dimorphic species studied. Taken together, the patterns across all species, and between different male dimorphic species, strongly support sexual selection on mating opportunities as the major determinant of male morph ratios in fig wasps.