Annually burned tallgrass prairie is purported to be a nitrogen-limited system, especially when compared to unburned prairie. To test the hypothesis that legumes, potential nitrogen-fixers, would increase in relative abundance in annually burned sites, we assessed their density and biomass for two seasons on upland and lowland soils in annually burned and unburned watersheds. Total legume density was significantly higher in burned (8.0 ± 1.0 [SE] stems/m2) than in unburned watersheds (3.0 ± 0.3 stems/m2). Species with higher (P < 0.05) densities in burned than in unburned prairie included Amorpha canescens, Dalea candida, Dalea purpurea, Lespedeza violacea, Psoralea tenuiflora, and Schrankia nuttallii. Desmodium illinoense was the only legume that responded negatively to annual fire. Total legume biomass did not differ between burned (11.3 ± 1.3 g/m2) and unburned prairie (10.5 ± 0.9 g/m2). Biomass productions of Dalea candida and Psoralea tenuiflora were higher (P < 0.05) in burned than in unburned sites, but biomasses of other legumes were similar between burn treatments. Average individual stem masses of Amorpha canescens and Baptisia bracteata were significantly greater in unburned than in burned prairie. Legumes were affected differentially by topographic location. Total legume density was higher (P < 0.05) on lowland soils (6.6 ± 1.0 stems/m2) than on upland soils (4.3 ± 0.5 stems/m2). However, total legume biomass was not different between lowland soils (12.0 ± 1.2 g/m2) and upland soils (9.9 ± 1.0 g/m2). Densities and biomasses of Amorpha canescens, Desmodium illinoense, and Lespedeza capitata were higher on lowland sites than on upland sites, whereas densities and biomasses of Baptisia bracteata and Dalea purpurea were higher on upland than on lowland soils. Most legume species are either fire tolerant or exhibit a positive response to fire and their persistence in annually burned prairie suggests that they may play an important role in the nitrogen budget of this ecosystem.