The potential for large-scale biodiversity losses as a result of climate change and human impact presents major challenges for ecology and conservation science. Governments around the world have established national parks and wildlife reserves to help protect biodiversity, but there are few studies on the long-term consequences of this strategy. We use Kenya as a case study to investigate species richness and other attributes of mammal communities in 6 protected areas over the past century. Museum records from African expeditions that comprehensively sampled mammals from these same areas in the early 1900's provide a baseline for evaluating changes in species richness and community structure over time. We compare species lists assembled from archived specimens (1896-1950) to those of corresponding modern protected areas (1950-2013). Species richness in Kenya was stable or increased at 5 out of 6 sites from historical to modern times. Beta-diversity, in contrast, decreased across all sites. Potential biases such as variable historical vs. modern collection effort and detection of small-bodied, rare, and low-visibility species do not account for the observed results. We attribute the pattern of decreased beta diversity primarily to increased site occupancy by common species across all body size classes. Despite a decrease in land area available to wildlife, our data do not show the extinctions predicted by species-area relationships. Moreover, the results indicate that species-area curves based solely on protected areas could underestimate diversity because they do not account for mammal species whose ranges extend beyond protected area boundaries. We conclude that the 6 protected areas have been effective in preserving species richness in spite of continuing conversion of wild grasslands to cropland, but the overall decrease in beta diversity indicates a decline in the uniqueness of mammal communities that historically characterized Kenya's varied landscape.