Abstract: Full use of the peak-shaving capacity of hydropower stations could bring more economic benefit. This paper describes a long-term optimization model that maximizes the peak-shaving benefit expectation based on a scenario reduction technique, the curve of peak-shaving price, and the otherness of daily load curves that reflects peak-shaving demands in different dispatch periods. This model extracts typical daily curves of peak-shaving operation for single-peak, double-peak and monthly-peak modes through simultaneous backward reduction, then simulates the curves of peak-shaving price for the three modes based on the daily curves extracted, and finally generates a peak-shaving price series. Calculations in a case study of the Three Gorges cascade hydropower stations show that the model could reflect the influence of different short-term peak-shaving operation modes on long-term dispatch and improve peak-shaving efficiency significantly, thus helping design long-term dispatch schemes of cascade hydropower stations.