Some recent models suggest a new role for evolutionary arms races between males and females in sexual selection. Female resistance to males is proposed to be driven by the direct advantage to the female of avoiding male-imposed reductions in the number of offspring she can produce, rather than by the indirect advantage of selecting among possible sires for her offspring, as in some traditional models of sexual selection by female choice. This article uses the massive but hitherto under-utilized taxonomic literature on genitalic evolution to test, in a two-step process, whether such new models of arms races between males and females have been responsible for rapid divergent evolution of male genitalia. The test revolves around the prediction that ‘new arms races’ are less likely to occur in species in which females are largely or completely protected from unwanted sexual attentions from males (e.g. species which mate in leks or in male swarms, in which males attract females from a distance, or in which females initiate contact by attracting males from a distance).

The multiple possible mechanical functions of male genitalia are summarized, and functions of male genitalic structures in 43 species in 21 families of Diptera are compiled. Functions associated with intromission and insemination (e.g. seizing and positioning the female appropriately, pushing past possible barriers within the female, orienting within the female to achieve sperm transfer), which are unlikely to be involved in new arms races when females are protected, are shown to be common (>50% of documented cases). This information is then used to generate the new arms race prediction: differences in genitalic form among congeneric species in which females are protected should be less common than differences among congeneric species in which females are vulnerable to harassment by males. This prediction was tested using a sample of 361 genera of insects and spiders. The prediction clearly failed, even when the data were adjusted to take into account several possible biases. Comparative analyses within particular taxonomic groups also failed to show the predicted trends, as did less extensive data on other non-genitalic male display traits. Arms races, as defined in some recent models, seem to have been less important in male–female coevolution of genitalic structures than has been suggested. By elimination, alternative interpretations, such as traditional female choice, which do not predict associations between female protection from harassment and rapid divergent evolution, are strengthened.