There is a general consensus that the diversity of a biotic community can have an influence on its stability, but the strength, ubiquity, and relative importance of this effect is less clear. In the context of biological invasions, diversity has usually been studied in terms of its effect on a community's invasibility, but diversity may also influence stability by affecting the magnitude of compositional or functional changes experienced by a community upon invasion. We examined the impacts of invasive ants on arthropod communities at five natural area sites in the Hawaiian Islands, and assessed whether differences among sites in community diversity and density variables were related to measures of stability. Ant invasion was usually associated with significant changes in overall community composition, as measured by Bray-Curtis distances, particularly among endemic subsets of the communities. Changes in mean species richness were also strong at three of the five sites. Among sites, diversity was negatively related to stability as measured by resistance to overall compositional change, but this effect could not be separated from the strong negative effect of invasive ant density on compositional stability. When compositional stability was measured as proportional change in richness, the best predictor of stability among endemic community subsets was endemic richness, with richer communities losing proportionately more species than species-poor communities. This effect was highly significant even after controlling for differences in invasive ant density and suggested that communities that had already lost many endemic species were resistant to further species loss upon ant invasion, while more intact communities remained vulnerable to species loss. Communities underwent strong but idiosyncratic functional shifts in association with ant invasion, both in terms of trophic structure and total arthropod biomass. There were no apparent relationships, however, between functional stability and community diversity or density measures. Instead, invasive ant density was the best among-site predictor of the magnitude of functional change. Overall, diversity appeared to be a poor predictor of stability in the face of ant invasion in these communities, possibly because any actual diversity effects were overshadowed by community-specific factors and variation in the magnitude of the ant-mediated perturbation.