A novel uncertainty model for short-term optimization scheduling of the hydrothermal power system based on chance-constrained programming is presented. It is taken into account that the relationship between day-ahead trading and real-time trading， constraints depending on hydrological and electric conditions， and uncertainty in hydrothermal power system. The constraints can be violated in some exceptional cases， but the probability of such a violation must be less than a specified level. And the cost of the day-ahead trading and real-time trading are considered in an integrated fashion for minimizing both the cost of thermal generation and power regulation for real-time balance during a given planning horizon. The model presented is solved by the combination of particle swarm optimization（PSO） and Monte Carlo simulation. A numerical example is served for demonstrating the feasibility of the method developed.