Although most efforts to improve habitat for bees and other pollinators focus on floral resources, it is also important to consider the availability and suitability of nesting habitats in relation to various management activities. We used soil emergence traps to compare the number of ground‐nesting bees among treatments in a long‐term fire frequency study in Florida. We regularly moved the traps, a strategy that appears to greatly improve the efficacy of this sampling approach. The experiment consisted of forested plots that have been burned annually, biennially or triennially for the past six decades as well as unburned reference plots. Bee abundance was significantly lower in the unburned plots than in any of the burn treatments and was higher in annual plots than triennial plots. Bee richness was significantly higher in plots burned annually than in unburned plots. Bee Shannon's diversity was significantly higher in plots burned biennially or triennially than in unuburned plots. Evenness was significantly lower in annually burned plots than in any of the other treatments. Indicator species analysis found six bee species to be significantly associated with one or more burn treatments. Four of them were strongly associated with plots burned annually or biennially while two species were significantly associated with plots burned biennially or triennially. None was significantly associated with unburned plots. Our findings suggest that frequent fires generally improve habitat for ground‐nesting bees in southeastern U.S. forests and that tolerances for particular burn frequencies vary by species.