The purpose of this study is to explore which model of Differential Evolution Algorithm (DE) is suitable for solving Vehicle Routing Problem (VRP). To improve the control parameters of the algorithm to enhance the solving ability of the algorithm, a full factor experiment and Kriging method are applied. The 15 test problems solved by the algorithm are downloaded from the NEO website. The solving ability of the algorithm is defined as the average of percentages of differences between the solved shortest distances and the true shortest distances. The results of this study have shown that DE/best/1/bin is most suitable for solving vehicle routing problems. In order to improve the control parameters (scaling factor F and crossover probability Cr) of DE/best/1/bin, a full factorial experiment is applied firstly to find the best level of parameters. Then, Kriging method is applied to convert the discrete experimental data into a continuous objective function. The optimization method is applied to find the best solution of parameters. Before improving the control parameters, the average of percentages of differences was as high as 46.27%. After applying the full-factorial experiment, the average of percentages of differences can be reduced to 24.46%. After applying the Kriging method, the average of percentages of differences can be reduced to 21.23%. When using DE/best/1/bin differential evolution algorithm to solve the vehicle routing problem, the optimal setting of the scaling factor F is 0.475259, and the optimal setting of the crossover probability Cr is 0.60151.