► Mixture yield exceeded 16tha−1 after four years with no fertilization or irrigation. ► A majority of multi-species plots exhibited overyielding. ► Moderately diverse prairie plant mixtures can improve forage productivity with minimal inputs. Ten prairie plant species were grown in monoculture and mixtures of two, four, six, and all ten species for four years at a study site near Blacksburg, VA (USA) to examine relationships between species richness of perennial prairie plant mixtures and forage yield. Mixtures were highly productive, exceeding 16t (1t=1000kg)ha−1 after four years with no fertilization or irrigation. Forage yield was affected by sown species richness only in years 1 and 4 when most mixture treatments yielded more than monoculture plots. The majority of multi-species plots (71%) exhibited a positive biodiversity effect where mixtures yielded more than respective monocultures. The data also suggested a strengthening of biodiversity effects with time (more multispecies plots with relative yield totals >1 and a positive net biodiversity effect). Mixtures that were more productive than the best performing species grown in monoculture (transgressive overyielding) increased from 25% in year 1 to 54% by year 4. Plots sown with ten species produced the most consistently high and stable yields and also effectively suppressed weeds. Livestock producers in cool-season regions could improve forage productivity with minimal external inputs by adding pastures sown with moderately diverse (n=4–10 species) prairie plant mixtures.