The relative importance of seed availability versus biotic interactions that affect early life stages in limiting plant population sizes and determining composition of plant communities is a central debate in plant ecology. We conducted a seed addition experiment in restored tallgrass prairie in central Kansas to determine (1) whether addition of seed of 18 native forb species produced persistent (three growing seasons) increases in the species' population sizes and plant species richness, (2) what properties of recipient communities best explained spatial variation in added species' establishment, and (3) whether seed size explained interspecific patterns in establishment success. Adding seed led to persistent increases in the number of added species present and in plant species richness at one of three sites. Increased species richness at the one site where community composition was structured by seed availability largely resulted from greater densities of four species. Seed size did not predict species' establishment success. Pre-existing plant species richness was correlated with added species' establishment success, but the direction of the relationship (positive vs. negative) varied among sites. Living aboveground plant biomass in experimental plots in the year of seed addition was negatively correlated with the number of added species established three years later. Our results provide further evidence for large spatial variation in seed limitation of plant community composition. Surprisingly, mean light availability and heterogeneity in light, both important parameters in conceptual models of grassland plant coexistence, did not predict the response of the recipient plant community to seed addition as well as pre-existing plant species richness and living aboveground biomass.