Sorghum (Sorghum bicolor L.) is well-known to adapt to stressful environmental conditions. However, yield variability of sorghum has not been thoroughly investigated for different growing conditions in a subtropical climate. The overall goal of this study was to identify yield potential and management options for sorghum production in the southeastern United States. Specific objectives were to calibrate and evaluate the Cropping System Model (CSM)-CERES-Sorghum model with sorghum variety trial data and to apply the model for determining hybrid performance as a function of sowing date for different environments in Georgia. The model was calibrated and evaluated with data for six grain sorghum hybrids from 31 variety trials conducted in Georgia and Florida. Following calibration and evaluation, the model was used to simulate grain yield of the six hybrids in response to seven sowing dates under rainfed and irrigated conditions at 10 locations in Georgia using 40 yr of historical weather data. The results indicate that normalized root mean square error (RMSEn) between simulated and observed yield for the six hybrids ranged from 1.4 to 19% for model calibration and from 12 to 28% for model evaluation, suggesting that the model can be calibrated and evaluated using limited data from variety trials. For the long-term scenarios, differences in simulated grain yield were found for hybrids, sowing dates, and locations under both rainfed and irrigated conditions. This demonstrated that the CSM-CERES-Sorghum model can be used to investigate the effects of climate variability on crop yield and to develop management practices for optimizing sorghum production.