This study was conducted to evaluate the CSM-CERES-Maize model for simulating maize growth and yield against different irrigation regimes in semi-arid conditions of Punjab (Pakistan). The data required to run the model were collected from field experiments conducted at University of Agriculture Faisalabad during 2009 and 2010. Two autumn-sown maize hybrids (Monsanto-919 and Pioneer-30Y87) were grown under variable irrigation regimes: I1 = no irrigation (control), I2 (4 irrigations, 1st at V6 i.e., 6 leaf stage, 2nd at V10, 3rd at V14, 4th at V18), I3 (6 irrigations, the first 4 as in I2 and the subsequent two at reproductive stages R1 and R2), I4 (8 irrigations, the first 6 as in I3 and the subsequent two at R3 and R4), I5 (10 irrigations, the first 8 as in I4 and the subsequent two at R5 and R6), I6: irrigation at 25 mm potential soil moisture deficit (PSMD) and I7: irrigation at 50 mm PSMD. The genetic co-efficients calibrated against the treatment I6 giving superior performance regarding yield and yield components for the two hybrids during 2009, were used to evaluate the model accuracy for irrigation treatments of 2010 season. The model performance during calibration was quite good. During evaluation, the model showed variability of 1.1 to 11.2% and -4.3 to 6.4% in total dry matter and -8.5 to 24.3% and -15.8 to 28.7% in grain yield for Monsanto-919 and Pioneer-30Y87, respectively for different irrigation regimes. Daily simulations of total dry matter were also good with d-statistics values higher than 0.970 in all treatments. The results showed that CSM-CERES-Maize can be successfully used in simulation studies of autumn-sown maize under semi-arid conditions of Punjab.