Fire, grazing, weather and associated changes in vegetation and environmental conditions affect small mammals in native tallgrass prairie in the Flint Hills of Kansas. In ungrazed sites, presence or absence of recurring fires influences plant production, structure of live and standing dead vegetation and density of litter (fallen and lodged plant debris). Based on studies on the Konza Prairie, most small mammals (i.e., rodents and shrews) respond either positively or negatively to fire-induced changes, which should change community structure and composition. We examined community characteristics (i.e., community abundance and evenness, species richness and diversity, community composition and inter-treatment similarity of communities) in annually burned (001A) and unburned treatments (020A) for 2 years and then changes that occurred when the fire regimes were reversed (treatments then renamed to reflect the new management regimen to R20A and R01A, respectively) over the subsequent 10-year period. Community abundance varied widely among years in both reversal treatments and showed no significant directional temporal change. Species richness varied from one to eight species across all treatments, seasons and years and was associated positively with community abundance. Despite this variability, species richness increased significantly in R20A in autumn over the 12-year study. Community evenness was positively associated (curvilinear patterns) with time period in both R01A and R20A in autumn and R01A in spring but it only approached statistical significance in R20A in spring. Likewise, species diversity mimicked the curvilinear patterns for evenness for both autumn and spring. The white-footed mouse (Peromyscus leucopus) initially was the largest proportional component (>0.8 of the community) in R01A in both autumn and spring but its proportional dominance significantly decreased through time. The only other species in R01A to have a value >0.5 was the hispid cotton rat (Sigmodon hispidus) in autumn. For R20A, the deer mouse (P. maniculatus) initially was the largest proportional component (>0.7 of the community) in both autumn and spring but its proportional dominance in autumn and spring significantly decreased through time to <0.20 of the community in autumn towards the end of the study. In contrast, three fire-negative species–the Elliot's short-tailed shrew (Blarina hylophaga), prairie vole (Microtus ochrogaster) and white-footed mouse–significantly increased their proportional abundances through time in autumn, although all species were <0.30 of the small mammal community. Communities in both treatments showed very low similarity values at the beginning of the study because one species of Peromyscus (although different species) numerically dominated in each of the two treatments. In both autumn and spring, similarity values significantly increased through time in a curvilinear pattern; spring communities were more similar than those in autumn. Our 12-year study was not long enough to observe the increase, peak and then decline in similarity that was expected.