This study investigated the relationship among plant diversity, production, stability, and susceptibility to invasion in restored northern tall grass prairies (United States). The experiment consisted of 50 species mixtures fertilized with N or P, at high or low levels. Results from the past 5 years were as follows: (1) aboveground biomass increased and year-to-year variability declined with increases in plant species and functional form richness, mostly as a result of substantial increases in minimum biomass (maximum biomass was unaffected). (2) Aboveground biomass and biomass stability increased when the species in the mixture had: (a) high relative growth rates, root density, root surface area per unit of root biomass, uptake rates of N or P per unit of root surface area, and N use efficiency and (b) low root to shoot ratio. (3) Invasion of nonseeded species declined with increases in plant species and functional form richness. (4) The results from this experiment did not provide a single specific criterion for selecting an optimal species mixture. However, if the objectives of the restoration were simply to achieve an aboveground biomass variability that is less than that of growing-season precipitation, then the seed mixtures need to have a minimum of nine species and three functional forms.