Physiological responses of six sorghum (Sorghum bicolor (L.) Moench) hybrids and their respective parental lines were evaluated under high and low soil moisture conditions at Tucson, Arizona in 1980 and 1981. Apparent photosynthesis, transpiration, diffusive resistance, and temperature differential (ambient temperature minus leaf temperature) were measured under field conditions at weekly intervals. To measure apparent photosynthesis, a small plexiglas chamber was sealed over a section of leaf blade and gas was sampled with two syringes pulled at a 30 or 60 second interval. The gas samples were injected from the syringes into an infrared gas analyzer to measure CO(,2) concentrations. Transpiration, diffusive resistance, and temperature differential were measured with a steady state porometer. Regression analysis was used to compare the physiological performance of the germplasm sources over a wide range of environmental conditions. The physiological characteristics of the highest yielding sorghum hybrid in 1980 were the most stable across all environments. This hybrid was superior to both its male and female parent for all four physiological characteristics. In 1980, this hybrid was superior to the other hybrids in temperature differential and transpiration. The superior yielding hybrid had the highest mean apparent photosynthesis and the lowest mean diffusive resistance. All hybrids had higher yields than their respective male parents under both irrigation treatments. Thirty-five days after planting, the superior yielding hybrid had the greatest leaf, stem, and root dry weights. This hybrid also had the highest relative leaf area expansion rate. Heterosis for stomatal density was not observed either year. Regression analysis proved to be an effective tool for analyzing the sorghum germplasm over a diverse range of environmental conditions. By comparing the response of an individual germplasm against the mean response of the population under numerous environmental conditions, the slope, mean, and coefficient of correlation may be used to evaluate genotype-environment interactions.