Three plant species from each of three grassland ecosystems were grown under elevated (700 mL/m3) and ambient (350 mL/m3) CO2 and were defoliated or left undefoliated to test whether species response to elevated CO2 and grazing is related to evolutionary grazing history or to mode of photosynthesis. The three ecosystems represented a tropical grassland dominated by C4 species (the Serengeti of Africa), a temperate grassland dominated by a mixture of C3 and C4 species (Flooding Pampa of South America), and a northern temperate grassland dominated by C3 species (Yellowstone National Park of North America). Plants were grown in growth chambers under common conditions to compare relative responses to grazing and elevated CO2. Elevated CO2 caused an increase in total biomass and total productivity (biomass 1 clippings) only in Yellowstone species, and increases in growth occurred primarily in crowns and roots (storage organs). There were no significant CO2 effects on biomass or productivity in Serengeti or Flooding Pampa species, and no CO2 effects on aboveground biomass or productivity (aboveground biomass 1 clippings) in species from any of the three ecosystems. Since aboveground plant parts are the portions that are available to grazing mammals, this suggests that increased atmospheric CO2 may not affect food quantity in these three grasslands. There was no interaction between CO2 and defoliation for any species; thus, it appears that herbivores will not affect how grasses respond to elevated CO2 (at least under average nutrient conditions). Elevated CO2 caused a reduction in leaf percentage of N in species from Yellowstone and Flooding Pampa (especially the C3 species, Briza subaristata), but not in Serengeti species. Because the quantity of food was unaffected by the CO2 treatments and forage N was reduced, grazing mammals in Yellowstone (elk, Cervus elaphus, and bison, Bison bison) and the Flooding Pampa (cattle) may be negatively affected. Responses to defoliation were fairly consistent among ecosystems in aboveground productivity, which did not differ between defoliated and undefoliated plants, and in leaf water potentials and percentage of N, both of which increased in response to defoliation. However, differences among ecosystems were found for crown and root biomass in response to defoliation: Serengeti species, on average, had higher crown and similar root biomasses after defoliation, whereas defoliated species from the other two ecosystems had reduced crown and root biomass.We suggest that the lower intensity and increased temporal variance in grazing pressure in Yellowstone vs. the Serengeti, selected for plants that shift allocation away from roots and crowns in order to compensate for aboveground herbivory.