Aiming at multi-objective Vehicle Routing Problem （VRP） with time windows， a Differential Evolution algorithm integrating Mutation Strategy and Adjacency Information （DE-MSAI） was proposed. Firstly， four mutation operators were designed by employing elite sampling strategy， so as to increase the algorithm search breadth. Secondly， customer adjacency information matrix was combined to guide the neighborhood search of the individuals， thereby improving local optimization efficiency. Finally， simulated annealing criterion was adopted to accept inferior solutions with certain probability. If the number of Pareto non-dominated solution set remained unimproved was beyond the threshold during optimization， elite fragment protection strategy would be activated to perturb a randomly selected solution from non-dominated solution set， thereby maintaining the population diversity. Simulation results based on Solomon standard library instances show that the proposed algorithm controls the solving error within 0.07% compared to Hybrid Crow Search Algorithm （HCSA）， and outperforms K-means clustering algorithm and Improved Large Neighborhood Search Algorithm （K-means-ILNSA） in most cases， achieving an average reduction of 4.51% in route deviation metric， verifying the effectiveness of the algorithm.