As cities expand, conservation of beneficial insects is essential to maintaining robust urban ecosystem services such as pollination. Urban warming alters insect physiology, fitness, and abundance, but the effect of urban warming on pollinator communities has not been investigated. We sampled bees at 18 sites encompassing an urban warming mosaic within Raleigh, NC, USA. We quantified habitat variables at all sites by measuring air temperature, percent impervious surface (on local and landscape scales), floral density, and floral diversity. We tested the hypothesis that urban bee community structure depends on temperature. We also conducted model selection to determine whether temperature was among the most important predictors of urban bee community structure. Finally, we asked whether bee responses to temperature or impervious surface depended on bee functional traits. Bee abundance declined by about 41% per °C urban warming, and temperature was among the best predictors of bee abundance and community composition. Local impervious surface and floral density were also important predictors of bee abundance, although only large bees appeared to benefit from high floral density. Bee species richness increased with floral density regardless of bee size, and bee responses to urban habitat variables were independent of other life-history traits. Although we document benefits of high floral density, simply adding flowers to otherwise hot, impervious sites is unlikely to restore the entire urban pollinator community since floral resources benefit large bees more than small bees.