After the status is analyzed on Wujiang River hydropower stations, a model of the mid-long term cascade hydropower station reservoir optimal operation is put forward and using Chaos Optimization Algorithm (COA) to solve the mid-long term cascade hydropower station reservoir optimal regulation problem. The main principle of article is using the randomness of chaos movement. First random chaos series are produced by Logistic mapping, career random series to feasible field S which include hydropower station’s objective function, by using of properties of randomness, periodicity and regularity to search optimal solution in global space, it can obtain point belong to the feasible field S. Then by comparison, iteration and secondary career wave, the paper get optimal scheduling graph of cascade hydropower station reservoir optimal regulation. Through the example of the scheduling model is validated, and the result shows that COA can solve nonlinear cascade hydropower station reservoir optimal regulation problem, which has complex constrain conditions. This algorithm not only makes the solution more accurate, converges faster, but also is an effective way to solve the problem of cascade hydropower station reservoir optimal regulation.