We investigated the e ects of elevated CO 2 (600 ll l A1 vs 350 ll l A1 ) and phosphorus supply (1 g P m A2 year A1 vs unfertilized) on intact monoliths from species-rich calcareous grassland in a greenhouse. Aboveground community dry mass remained almost una ected by elevated CO 2 in the ®rst year (+6%, n.s.), but was signi®cantly stimulated by CO 2 enrichment in year two (+26%, P<0.01). Among functional groups, only graminoids contributed signi®cantly to this increase. The e ect of phosphorus alone on community biomass was small in both years and marginally signi®cant only when analyzed with MANOVA (+6% in year one, +9% in year two, 0.1 ³ P > 0.05). Belowground biomass and stubble after two seasons were not di erent in elevated CO 2 and when P was added. The small initial increase in aboveground community biomass under elevated CO 2 is explained by the fact that some species, in particular Carex¯acca, responded very positively right from the beginning, while others, especially the dominant Bromus erectus, responded negatively to CO 2 enrichment. Shifts in community composition towards more responsive species explain the much larger CO 2 response in the second year. These shifts, i.e., a decline in xerophytic elements (B. erectus) and an increase in mesophytic grasses and legumes occurred independently of treatments in all monoliths but were accelerated signi®cantly by elevated CO 2 . The difference in average biomass production at elevated compared to ambient CO 2 was higher when P was supplied (at the community level the CO 2 response was enhanced from 20% to 33% when P was added, in graminoids from 17% to 27%, in legumes from 4% to 60%, and in C.¯acca from 120% to 298% by year two). Based on observations in this and similar studies, we suggest that interactions between CO 2 concentration, species presence, and nutrient availability will govern community responses to elevated CO 2 .