This paper was focused on the network design problem of the reverse logistics and presented a multi-period model for the location of distribution and recycling centers in a reverse logistics network. Taking into account the uncertainty of recovery products quantity and the stochastic demands of remanufacturing products, this paper proposed a dynamic location model of remanufacturing reverse logistics network based on time series smoothing method and dynamic programming. It was trying to introduce the customer demands, uncertain factors, and the cost of the whole supply chain into the model. Instead of traditional cost saving objective, this research proposed a new model with the object function of maximizing the total profit. The time series smoothing method was used to predict the customer demands. And the dynamic programming was introduced to solve the problem.