Abstract: Various stochastic factors have mild time-varying negative impact on the flood control safety due to the ageing of hydraulic projects.It is important that the time-varying characteristics of the various stochastic factors are quantitatively predicted to evaluate the time-varying flood control risk rate.A frame to infer the statistical distributions of the stochastic variables is provided by the Bayesian approach.By using the prior information the results of suitable sampling inspections can be used to modify and improve the hypothesis of accumulated statistical distribution to control the time-varying uncertainty.The time-varying characteristics of seepage variable L were analyzed,the feasibility and suitability by using the Bayesian approach to quantitatively predict the time-varying stochastic characteristics were proved.The inspection procedure simulated by using the Normal conjugate function was suggested,and the posterior distribution of L was computed.On this basis,a practical model for quantitatively predicting the seepage risk rate of the embankments and dams was discussed,and the influence of the time-varying characteristics of the stochastic variable L on seepage risk rate was analyzed.