Most studies of extinction selectivity have focused on mass extinctions. Here we analyze the patterns of susceptibility to extinction during the late Pliocene and Pleistocene of California in the family Pectinidae. The Pectinidae declined in diversity from a high of 32 species in the late Pliocene to the current level of 11 species living in the California region, indicating that the composition of the living fauna has been shaped primarily by extinction in the last several million years. Extinction appears to have occurred in one large pulse, but because of uneven sampling we were unable to resolve the timing further and have analyzed the patterns of extinction treating the late Pliocene through middle Pleistocene as a single period of elevated extinction. Extinctions were not random with regard to taxonomic relationships. Species-level extinctions were higher in more speciose genera, but these genera were buffered against genus-level extinctions. This resulted in a disproportionately large number of monospecific genera in the living fauna. In addition, extinctions were not random with regard to body size, with large species preferentially surviving. This selectivity pattern is evident only when clade membership is taken into account; when analyzed across the entire family, no pattern of size selectivity was apparent. Our results suggest that (1) patterns of extinction selectivity at the genus level may be a poor proxy for species-level patterns, and (2) whole-fauna analyses may not uncover strong selectivity within lineages.