The ecological role of biodiversity in achieving successful restoration has been little explored in restoration ecology. We tested the prediction that we are more likely to create persistent, species-rich plant communities by increasing the number of species sown, and, to some degree, by varying functional group representation, in experimental prairie plantings. There were 12 treatments consisting of 1-, 2-, 3-, 4-, 8-, 12-, and 16-species mixtures of native perennials representing four functional groups (C4 grasses, C3 grasses, nitrogen-fixing species, and late-flowering composites) that predominate within Central Plains tallgrass prairies. In 2000, species were seeded into square plots (6 × 6 m), with five replicates per treatment, on former agricultural land. Annually, we measured total species richness and evenness, target species richness and cover, and richness and cover of resident species (i.e., those emerging from the seed bank). Both target species richness and rate of establishment of target communities were highest in the most species-rich mixtures, but there was no additional benefit for treatments that contained more than eight species. Richness of resident species did not vary with target species richness; however, cover by resident species was lower in the higher target species treatments. Our results, indicating that establishment of species-rich prairie mimics can be enhanced by starting with larger numbers of species at the outset, have implications for grassland restoration in which community biodiversity creation and maintenance are key goals.