The present study aimed to calibrate, validate, and conduct a sensitivity analysis of the AquaCrop crop model for simulating yield and irrigation water use efficiency of roselle under different irrigation and potassium fertilizer levels. The field experiment was conducted as a split-plot design based on a randomized complete block design with three replications in , Sistan and Baluchestan, Iran, during 2018 and 2019. Treatments consisted of four irrigation levels corresponding to 40, 60, 80, and 100% of crop water requirement and three potassium fertilizer levels corresponding to 50, 75, and 100 of the recommended potassium requirement. The sensitivity of the model to the parameters WP*, KcTrx, CC₀, CGC, CDC, and HI was evaluated using the method proposed by Beven (1979). During calibration, the values of R², RMSE, and RE for yield were 0.99, 25.17 kg ha⁻¹, and 5.36%, respectively, while the corresponding values for irrigation water use efficiency were 0.96, 0.05 kg m⁻³, and 10.21%, respectively. During validation, these values were 0.97, 29.14 kg ha⁻¹, and 7.51% for yield and 0.95, 0.07 kg m⁻³, and 14.29% for irrigation water use efficiency, respectively. The sensitivity coefficients for WP*, HI, KcTrx, CGC, CC₀, and CDC ranged from 0.74 to 0.95, 0.67 to 0.87, 0.53 to 0.81, 0.38 to 0.57, 0.032 to 0.084, and −0.063 to −0.043, respectively. The AquaCrop showed the highest sensitivity to WP*, followed by HI and KcTrx. The calibration and validation results also indicated that the model provided satisfactory accuracy in simulating roselle yield and irrigation water use efficiency.