Abstract: The simulation of crop yield is of great significance to develop irrigation scheduling and planting patterns, in order to ensure water and food security in the world. The AquaCrop-KR model has been commonly used as the non-linear equation to fit the relationship between the aboveground biomass and crop transpiration, as well as the water production functions. The harvest index was simulated for the higher prediction accuracy of the crop yield under different water regimes. However, the planting density cannot be considered in this model. The objective of this study was to modify the AquaCrop-KR model, and then simulate the hybrid maize seed yield under different water regimes and planting densities in an arid region of Northwest China. Two field experiments were conducted at the National Field Scientific Observation and Research Station on the Efficient Water Use of Oasis Agriculture in Wuwei of Gansu Province of China (37°52′N, 102°50′E) from 2013 to 2016. In the first experiment, six planting densities were set as 6.75, 8.25, 9.75, 11.25, 12.75, and 14.25 plants/m2 from 2013 to 2015. In the second experiment, there were 12 treatments in 2015, with three irrigation levels (full irrigation, 2/3 of full irrigation, and 1/3 of full irrigation) during the growing season, and four planting densities (8.25, 9.75, 11.25, and 12.75 plants/m2). Specifically, 1/3 of full irrigation was replaced by 1/2 of full irrigation in 2016. But, the rest of the irrigation levels were consistent with 2015. Some parameters were collected in both experiments, including the soil water content, evaporation, aboveground biomass, grain yield, and weather data. After that, the planting density factors were introduced to modify the normalized water productivity and harvest index in the AquaCrop-KR model. The calibration results showed that there was a parabolic relationship between harvest index and planting density, which first increased and then decreased. There was an increase in the water sensitivity indexes of harvest index at the vegetative, flowering, and reproductive stages, as the planting density increased. In addition, the normalized water productivity showed a unimodal change with first increased and then decreased with the increasing cumulative normalized crop transpiration. By contrast, the maximum of the normalized water productivity decreased with the increasing planting density, whereas, there was an increase in the corresponding cumulative normalized crop transpiration. The validation results showed that the modified AquaCrop-KR underestimated the grain yield by 5%, compared with the measurements, with the determination coefficient, relative root mean square error, average relative error, modeling efficiency, and agreement index were 0.87, 0.079, 0.057, 0.750, and 0.942, respectively. It infers that the modified model can be used to simulate the grain yield of hybrid maize. This finding can also provide a theoretical reference to predict the crop yield under different water regimes and planting densities.