This paper considers congestion road pricing with demand uncertainty，and uses the mean-excess total travel time as the system risk measure．Mean-excess total travel time can consider both the reliability and the unreliability of total travel time，and so it is a more complete risk measure．Because the model in this paper with stochastic demand is difficult to solve，we turn it to a deterministic model through Monte Carlo simulation，and solve it with the genetic algorithm．A numerical example is presented to illustrate the model and compare it with the expected value model．It shows our model is more appropriate for risk-aversion deciders．At last，we demonstrate the stability of the algorithm through examining the effect of the parameters to the solution．