Summary1. Models that predict organismal and population responses to climate change may be improved by considering ecological factors that affect species thermal tolerance. Species differences in microhabitat use can expose animals to diverse thermal selective environments at a given site and may cause sympatric species to evolve different thermal tolerances. 2. We tested the hypothesis that species differences in body size and microhabitat use (abovevs. below-ground activity) would correspond to differences in thermal tolerance (maximum critical temperatures: CT max ). Thermal buffering effects of soil can reduce exposure to extreme high temperatures for below-ground active species. We predicted larger-bodied individuals and species would have higher CT max and that species mean CT max would covary positively with degree of above-ground activity. We used Neotropical army ants (Formicidae: Ecitoninae) as models. Army ants vary in microhabitat use from largely subterranean to largely above-ground active species and are highly size polymorphic. 3. We collected data on above-and below-ground temperatures in habitats used by army ants to test for microhabitat temperature differences, and we conducted CT max assays for army ant species with varying degrees of surface activity and with different body sizes within and between species. We then tested whether microhabitat use was associated with species differences in CT max and whether microhabitat was a better predictor of CT max than body size for species that overlapped in size. 4. Microhabitat use was a highly significant predictor of species' upper thermal tolerance limits, both for raw data and after accounting for the effects of phylogeny. Below-ground species were more thermally sensitive, with lower maximum critical temperatures (CT max ). The smallest workers within each species were the least heat tolerant, but the magnitude of CT max change with body size was greater in below-ground species. Species-typical microhabitat was a stronger predictor of CT max than body size for species that overlapped in size. Compared to the soil surface, 10-cm subsoil was a significantly moderated thermal environment for below-ground army ants, while maximum surface raid temperatures sometimes exceeded CT max for the most thermally sensitive army ant castes. 5. We conclude sympatric species differences in thermal physiology correspond to microhabitat use. These patterns should be accounted for in models of species and community responses to thermal variation and climate change.