Based on the multiple dairy products distribution in Beijing, the optimization of location-routing for joint distribution was studied. A mixed integer programming model was formulated for the joint distribution location-routing problem with two-echelon capacitated constraints. A three-phase algorithm was designed for solving it. First, all customers were divided into multiple customer sets by K-means clustering method based on genetic algorithm. Then, the optimal delivery routes and costs for each candidate distribution center serving each customer set were calculated, by which the joint distribution centers location and the secondary level delivery routing problem were simplified to that of joint distribution centers location and customer sets allocation.The mathematical model was formulated, and solved by Lingo software. Finally, the first level delivery routes from the logistics center to the selected joint distribution centers were determined. By comparing the costs of separate distribution mode and joint distribution mode of two brands of dairy products in Beijing, the rationality and effectiveness of the model and algorithm were verified. It provides the decision-making basis for solving the problems such as the joint distribution network optimization of multi-brand daily products.