In a given watershed, the rainfall input has an important impact on distributed hydrological simulation. Thus, there is a real need for more rainfall gauges in order to reflect rainfall variability and its effect on runoff prediction in the watershed scale. However, the rainfall data may be incompleted in some watersheds. Aiming at the problem, this study interpolated the rainfall data with the inverse distance weighted method in the Wuliehe River Basin, and the soil and water assessment tool (SWAT) was applied to analyze the effect of different rainfall input on runoff simulation, and the measured data of rainfall gauges were used for SWAT calibration and validation. The results indicate that different rainfall input have less affect on the average areal rainfall, but the area rainfall would be severely underestimated based on the rainfall data from the weather stations in some years, and largely re - flected during the flood periods. The different rainfall input has significant effects on runoff prediction. Under the situation of rain - fall incompleteness, compared with the directly utility of rainfall data of weather stations within the study area, using the interpolated rainfall data of gauges can increase the accuracy of runoff simulation to some extent, and to fulfill the distributed hydrological simulation in the basins that rainfall data is incompleteness.