Despite being commonly used as systematic characters earlier this century, relatively few systematists currently use behavioral information for phylogenetic reconstruction. A natural experiment provided by the inferred phylogenetic relationships and host associations of a tribe of ant guest beetles allows for a novel test of the phylogenetic utility of behavior. Historical and current evolutionary forces have acted in opposition during recent behavioral evolution, making it possible to determine the relative evolutionary lability of two general classes of behavioral characters. I dissected the behavioral phenotype into two components and tested for the relative presence of these two opposing forces on each behavioral component. Behavior was broken up into the continuously differing behavioral component (durations of functionally similar beetle behaviors) and the discretely differing behavioral component (species-typical behaviors). Conducting a series of statistic and cladistic tests on each component separately, I concluded that current evolutionary forces primarily determined the continuous behavioral component, because it was most similar among heterogeneric, nestmate beetle species. In contrast, I concluded that ancient evolutionary forces determined the discrete component because it was most similar among congeneric, non-nestmate species. These results support two hypotheses. First, the durations for which behaviors are performed evolve quicker than do species-typical behavioral traits. Second, species-typical behavioral characters are often as nonhomoplastic for phylogenetic reconstruction as are morphological characters.