Neuropeptides coordinate complex social behaviors important to both basic and applied science. Understanding such phenomena requires supplementing the powerful tools of behavioral neuroscience with less conventional model species and more rigorous evolutionary analyses. We review studies that use comparative methods to examine the roles of vasopressin and oxytocin in mammalian social behavior. We find that oxytocin and vasopressin receptor distributions are remarkably variable within species. Studies of socially monogamous prairie voles reveal that pronounced individual differences in spatial memory structures (retrosplenial cortex and hippocampus) are better predictors of social and sexual fidelity than are areas known to regulate pairbonding directly, a pattern that seems to be mediated by the contributions of the neuropeptides to space use in natural settings. We next examine studies of individual and species differences in cis-regulatory regions of the avpr1a locus. While individual differences in social behaviors are linked to length of a microsatellite at the avpr1a locus, phylogenetic analyses reveal that the presence or absence of a microsatellite does not explain major differences between species. There seems to be no simple relationship between microsatellite length and behavior, but rather microsatellite length may be a marker for more subtle sequence differences between individuals. Lastly, we introduce the singing mouse, Scotinomys teguina, whose neuropeptide receptor distributions and unique natural history make it an exciting new model for mammalian vocalization and social cognition. The findings demonstrate how taxonomic and conceptual diversity provide a broader basis for understanding social behavior and its dysfunction.