To determine the long-term impact of elevated CO2 on primary production of native tallgrass prairie, we compared the responses of tallgrass prairie at ambient and twice-ambient atmospheric CO2 levels over an 8-year period. Plots in open-top chambers (4.5 m diameter) were exposed continuously (24 h) to ambient and elevated CO2 from early April to late October each year. Unchambered plots were monitored also. Above-ground peak biomass was determined by clipping each year in early August, and root growth was estimated by harvesting roots from root ingrowth bags. Plant community composition was censused each year in early June. In the last 2 years of the study, subplots were clipped on 1 June or 1 July, and regrowth was harvested on 1 October. Volumetric soil water content of the 0–100 cm soil layer was determined using neutron scattering, and was generally higher in elevated CO2 plots than ambient. Peak above-ground biomass was greater on elevated CO2 plots than ambient CO2 plots with or without chambers during years with significant plant water stress. Above-ground regrowth biomass was greater under elevated CO2 than under ambient CO2 in a year with late-season water stress, but did not differ in a wetter year. Root ingrowth biomass was also greater in elevated CO2 plots than ambient CO2 plots when water stress occurred during the growing season. The basal cover and relative amount of warm-season perennial grasses (C4) in the stand changed little during the 8-year period, but basal cover and relative amount of cool-season perennial grasses (C3) in the stand declined in the elevated CO2 plots and in ambient CO2 plots with chambers. Forbs (C3) and members of the Cyperaceae (C3) increased in basal cover and relative amount in the stand at elevated compared to ambient CO2. Greater biomass production under elevated CO2 in C4-dominated grasslands may lead to a greater carbon sequestration by those ecosystems and reduce peak atmospheric CO2 concentrations in the future.