To restore diversity of native vegetation, we must understand factors responsible for diversity in targeted communities. These factors operate at different spatial scales and may affect the number and relative abundances of species differently. We measured diversity of plant species and functional groups of species in replicated plots within paired restored and remnant (relic) tallgrass prairies at three locations in central Texas, U.S.A. To determine the contributions of species abundances and of spatial patterns of diversity to differences between prairie types, we separated diversity into richness and evenness (relative biomass) and into within-plot (α), among-plot (β), and prairie (γ) components. Species diversity was greater in remnant than in restored prairies at all spatial scales. At the γ scale, both species richness and species evenness were greater in remnants because of greater spatial variation in species composition. At the α scale, remnants were more diverse because of greater richness alone. Mean α richness correlated positively with the size of the species pool in restored prairies only, implying that in remnants, α richness was influenced more by colonization dynamics than by the number of species available for colonization. Plots in remnant prairies contained more functional groups and fewer species per group than did plots in restored prairies, suggesting that resource partitioning was greater in relic prairies. Our results are consistent with the interpretation that local ecological processes, like resource partitioning and limitations on seed dispersal, contribute to the greater diversity of remnant than restored prairies in central Texas. Restoration practices that limit abundances of competitive dominants, increase the number of species in seed mixtures, and increase the proximity of plants of different functional groups thus may be required to better simulate the plant diversity of tallgrass prairies.