Carbon dioxide effluxes from plants, litter and soil were measured in two mixed-grassland sites in Saskatchewan, Canada. Ecosystems at both locations were dominated by Agropyron dasystachyum (Hook.) Scribn. Respiration rates of intact and experimentally-modified systems were measured in field chambers using alkali-absorption. Removal of green leaves, dead leaves, and litter from a wet sward reduced respiration to as low as 58% of the rate in an intact system. In a dry sward green shoots were the only significant above-ground source of CO.Carbon dioxide effluxes from different parts of A. dasystachyum plants, and from soil samples were measured in laboratory vessels at 20Β° using alkali-absorption. Respiration of green leaves (1.46 mg CO g h) was significantly higher than microbial respiration in moist, dead leaf samples (0.79 mg CO g h) or litter (0.75 mg CO g h). Microbial respiration in air-dried, dead plant material was very low. Average repiration rates of roots separated from soil cores (0.24 mg CO g h) were lower than many values reported in the literature, probably because the root population sampled included inactive, suberized and senescent roots. Root respiration was estimated to be 17-26% of total CO efflux from intact cores.Laboratory data and field measurements of environmental conditions and plant biomass were combined in order to reconstruct the CO efflux from the shoot-root-soil system. Reconstructed rates were 1.3 to 2.3 times as large as field measured rates, apparently because of stimulation to respiration caused by the experimental manipulations. The standing dead and litter fractions contributed 26% and 23% of the total CO efflux in a wet sward. Both field-measured and reconstructed repiration values suggest that in situ decomposition of standing dead material under moist conditions can be a significant part of carbon balance in mixed grassland.