Abstract: The accurate prediction of reservoir inflow is the foundation of power generation in hydropower station. According to uncertainty of reservoir inflow, a stochastic simulation method is proposed in this paper. Reservoir inflow is treated as a random process, composite method of stepwise regression and autoregression is used to analyze trend term of time sequence, periodic term and stationary term of reservoir inflow, and random component is added to describe its random variation. Taking case of a hydropower station in Guangxi, it is verified that the combined model of stepwise regression and autoregression has a high fitting degree and feasibility to simulate random reservoir inflow sequence by comparing with historical real data.