Soil water deficits are likely to influence the response of crop growth and yield to changes in atmospheric CO2 concentrations (C-a), but the extent of this influence is uncertain. To study the interaction of water deficits and C-a on crop growth, the ecosystem simulation model ecosys was tested with data for diurnal gas exchange and seasonal wheat growth measured during 1993 under high and low irrigation at C-a = 370 and 550 mu mol mol(-1) in the Free Air CO2 Enrichment (FACE) experiment near Phoenix, AZ. The model, supported by the data from canopy gas exchange enclosures, indicated that under high irrigation canopy conductance (g(c)) at C-a = 550 mu mol mol(-1) was reduced to about 0.75 that at C-a = 370 mu mol mol(-1), but that under low irrigation, g(c) was reduced less. Consequently when C-a was increased from 370 to 550 mu mol mol(-1), canopy transpiration was reduced less, and net CO2 fixation was increased more, under low irrigation than under high irrigation. The simulated effects of C-a and irrigation on diurnal gas exchange were also apparent on seasonal water use and grain yield. Simulated vs. measured seasonal water use by wheat under high irrigation was reduced by 6% vs. 4% at C-a = 550 vs. 370 mu mol mol(-1), but that under low irrigation was increased by 3% vs. 5%. Simulated vs. measured grain yield of wheat under high irrigation was increased by 16% vs. 8%, but that under low irrigation was increased by 38% vs. 21%. In ecosys, the interaction between C-a and irrigation on diurnal gas exchange, and hence on seasonal crop growth and water use, was attributed to a convergence of simulated g(c) towards common values under both C-a as canopy turgor declined. This convergence caused transpiration to decrease comparatively less, but CO2 fixation to increase comparatively more, under high vs. low C-a. Convergence of g(c) was in turn attributed to improved turgor maintenance under elevated C-a caused by greater storage C concentrations in the leaves, and by greater rooting density in the soil.