Based on the principle for minimizing the total logistics cost, a location-allocation model of three-stage supply chain with multi-products is established. The model takes the transportation expenses of raw materials and products, the fixed construction cost, the cost of raw materials, and the processing cost of products for factories, as well as fixed construction fee and storage fee for sales centers into account. In the stage of supplier-factory, a factory not only can purchase different raw materials from different suppliers, but also can purchase the same kind of raw materials from different suppliers; however, in the stages of factory-sales center and sales center-customer, though a product demander can purchase different products from different suppliers, however the same kind of products must be purchased from a single supplier. Based on these characteristics of the model, the stochastic order encoding in a location problem is combined with the priority encoding in a transportation problem to design a new chromosome encoding method, the method is able to use genetic algorithm for solving the location-allocation problem of three stage supply chain with multi-products. The simulation results of numerical experiment show that the proposed encoding technique of chromosome is suitable, and the genetic algorithm has the fairly good research efficiency, it is able to rapidly and stably solve the location-allocation problem of three-stage supply chain with multi-products.