[Objective] The paper aims to evaluate the influence of changes in parameters of SWIM (soil and water integrated model) on the simulation accuracy of runoffs in Wangjiaba region, and provides scientific support for flood forecasting and disaster reduction, and provides a reference for the application of SWIM in other areas.[Methods] Based on the measured data at Wangjiaba region during the calibration period (1959-1978) and the validation period (1979-2008), the optimal parameter combination was obtained, and the sensitivity of SWIM to the seven calibratable parameters in Wangjiaba region was quantitatively analyzed.[Results] The Nash-Sutcliffe efficiency (NSE) coefficients of runoffs in the calibration and validation periods were 0.79 and 0.81, respectively, with relative errors of 22% and 7.8%, respectively. With the increase of the thc parameter (correction factor for potential evapotranspiration on sky emissivity), the relative error increased significantly, and its sensitivity to the thc value was higher than that of the NSE coefficient. roc₄ (river routing coefficient) was more effective than roc₂ (river routing coefficient), while roc₂ had a larger effect on runoff in flooding season than in non-flooding season. With the increase of the sccor parameter (correlation factor for saturated conductivity), the simulation accuracy increased during the flooding season and decreased during the non-flooding season.[Conclusion] The calibrated SWIM model shows good applicability in Wangjiaba region. The results of SWIM simulation in Wangjiaba region are sensitive to variations in the parameters thc, cnum₁ (CN condition 1 when the curve number method code is 1), cnum₃ (CN condition 3 when the curve number method code is 1), roc₂, roc₄ and sccor, but not to variations in the parameters bff (base flow factor), gwq₍₎ (contribution of initial groundwater flow to runoff) and abf₍₎ (groundwater alpha factor).