Abstract: Rainfall-runoff models are important for flood forecasting and control, but large errors in their predictions could be produced even using the data of in-situ rainfall observation due to uncertainties in their parameters. Therefore, certain risks exist in the flood control decision making that is based on the forecasting results. For a quantitative analysis of such risks, this paper describes a flood risk analysis method that integrates a generalized likelihood uncertainty estimation (GLUE) method and simulations of reservoir operation. By using a large number of equivalent parameters, this method calculates a set of reservoir inflows for possible flood processes and estimates their effects on scheduling decision so as to obtain the probability of risk events. Its application in a case study of the Shuangpai basin together with the Xin’anjiang model shows satisfactory effects in reducing forecast errors and advantages in the decision making of flood forecasting and control.