The reverse logistics responsible for collecting products from customers and return them to providers is increasingly receiving much attention in the era of E-commence. What's more, to reduce reverse logistic cost, the enterprise tends to outsourcing these activities to the third-party reverse logistics service providers (3PLs). Previous studies on the design of reverse logistics network are mainly based on traditional business and assume that the demand and the return are known. However, in this paper, we propose a decision reverse logistics network model under the context of E-business with the assumption of uncertain demand and return which are determined by the optimal return policy. The major focus of this model is on determining the optimal number and location of factories, on-line retails and the 3PLs. And finally, a numerical example is developed to illustrate the practicability and efficiency of the model.