Historical records of species are compared with current records to elucidate effects of recent climate change. However, confounding variables such as succession, land-use change, and species invasions make it difficult to demonstrate a causal link between changes in biota and changes in climate. Experiments that manipulate temperature can overcome this issue of attribution, but long-term impacts of warming are difficult to test directly. Here we combine historical and experimental data to explore effects of warming on ant assemblages in southeastern US. Observational data span a 35-year period (1976–2011), during which mean annual temperatures had an increasing trend. Mean summer temperatures in 2010–2011 were ∼2.7°C warmer than in 1976. Experimental data come from an ongoing study in the same region, for which temperatures have been increased ∼1.5–5.5°C above ambient from 2010 to 2012. Ant species richness and evenness decreased with warming under natural but not experimental warming. These discrepancies could have resulted from differences in timescales of warming, abiotic or biotic factors, or initial species pools. Species turnover tended to increase with temperature in observational and experimental datasets. At the species level, the observational and experimental datasets had four species in common, two of which exhibited consistent patterns between datasets. With natural and experimental warming, collections of the numerically dominant, thermophilic species, Crematogaster lineolata, increased roughly two-fold. Myrmecina americana, a relatively heat intolerant species, decreased with temperature in natural and experimental warming. In contrast, species in the Solenopsis molesta group did not show consistent responses to warming, and Temenothorax pergandei was rare across temperatures. Our results highlight the difficulty of interpreting community responses to warming based on historical records or experiments alone. Because some species showed consistent responses to warming based on thermal tolerances, understanding functional traits may prove useful in explaining responses of species to warming.