To optimize the loading of products in the process of order fulfillment is important to improve the efficiency of the logistics system of enterprises. Aiming at the three-dimensional multiple bin-size bin packing problem with large size differences and strong heterogeneity of products, the size classification strategy was proposed, and a hybrid heuristic algorithm was designed based on it. In the algorithm, firstly, the products were divided into different size types according to the matching degree of products and pallets, and the corresponding loading strategies were designed for different types of them. Then the best fit algorithm was used to select the appropriate cartons for the products. Finally, greedy random adaptive search procedure was used to find the packing plan. Taking the real order as the dataset, the test results show that the proposed algorithm is superior to the traditional three-dimensional packing algorithm, and the layout of the packing plan found by using the size classification strategy is more reasonable, which can provide decision support for similar packing problems.