Field experiments were conducted for two seasons in Ile-Ife, Nigeria to evaluate the performance of the AquaCrop model in simulating the effects of soil fertility management on the canopy cover (CC), soil water storages (SWS), cumulative aboveground biomass (BM), evapotranspiration (ETa), grain yields, and water productivity (WP) of rainfed maize. Six levels of soil fertility management and two cultivars of maize, SUWAN 1-SR and PVA led to a 2 by 6 factorial experimental treatment and arranged in a randomized complete block design. Agronomic and environmental parameters were measured for two consecutive seasons. The AquaCrop model was calibrated using data from the wetter year. The AquaCrop model captured well the variances in the CC, R2 ≥ 0.88, RMSE ≤ 14.2, and d-index ≥ 0.97 under full and stressed soil fertility. Although the AquaCrop model over and underestimated SWS, it is still within acceptable limits. The model simulated SWS well, R2 ≥ 0.71, EF ≥ 0.97, and d-index ≥ 0.97. AquaCrop tends to underestimate ETa under rainfall and NPK variabilities. The AquaCrop model simulated grain yields excellently, R2 = 0.99, b = 1.00. The 150% of the recommended NPK application is suitable for the desired improvement in land and water productivity of the crop. The AquaCrop model predicted and captured the trends in the yields and water productivity of maize adequately under varying NPK applications. Further research is required on other cultivars of the crop and locations in the area in order to generalize the adequacy of the model.