Abstract: The design of the rain gauge network affects the accuracy of model simulation. Therefore,studying the effect of rain gauge density and its distribution on improving runoff simulation accuracy and reducing the modeling uncertainty is of vital importance. In this paper,the Xin'anjiang model and the HBV model were applied to simulate the runoff of the Xiangjiang River basin,and the Bayesian method was used to analyze the runoff simulation uncertainty resulted from resampling of rain gauge networks with different gauge densities and spatial distributions. The results reveal that increasing the rain gauge density can reduce the estimation error of areal rainfall,which in turn,improves model simulation accuracy;optimizing the rain gauge number and location can reduce the uncertainty of areal mean rainfall,thereby improving the runoff simulation accuracy;under the same rainfall input and parameter sampling methods,the simulation uncertainty of the Xin'anjiang model and the HBV model has similar characteristics,but the overall simulation accuracy of the Xin'anjiang model is higher,and the uncertainty of the HBV model is bigger.