Alfalfa (Medicago sativa L.) and orchard grass (Dactylis glomerata L.) plots were exposed to ambient or ambient plus 350 cm(3) m(-3) carbon dioxide concentrations at Beltsville, Maryland, U.S.A. Replicate plots were established in different years and fertilized annually. We report here data for the second and third years after establishment. There has been no increase in the yearly production of either species at the elevated carbon dioxide concentration after the first season. In orchard grass, reduced growth at the high carbon dioxide concentration in the spring offset growth stimulation in the summer. Weed growth was consistently increased by carbon dioxide enrichment, but weed species composition was unaffected. Leaf photosynthetic capacity was reduced by the high carbon dioxide concentration in both crop species, as was leaf nitrogen content. Canopy carbon dioxide uptake was slightly higher in the elevated carbon dioxide treatments, consistent with the increased weed growth. In alfalfa, elevated carbon dioxide significantly reduced canopy carbon dioxide efflux at night for the same daytime uptake rate and temperature. The growth conversion efficiency estimated from elemental composition of the tissue was not substantially altered by carbon dioxide treatment in either crop species, indicating little effect of carbon dioxide treatment on the respiratory cost of tissue synthesis. Canopy conductance to water vapour averaged 23% less at high than at low carbon dioxide in the orchard grass plots, and 14% less in the alfalfa plots. This was consistent with the smaller short-term response of conductance to carbon dioxide concentration in the alfalfa plots. It is concluded that a warm climate and fertile soil does not guarantee a persistent response of production to elevated carbon dioxide concentration in these herbaceous perennial species.