The vehicle routing problem whose solution is a key to improve efficiency of logistics problem is a classical NP - hard problem,and it is usually difficult for traditional methods to obtain satisfying solutions so as to high logistics costing. In this paper, in order to reduce logistics costs, the hybrid genetic algorithm was selected to solve the VRP problem. This paper established the VRP mathematic model at first. Second, the improvement using Greedy Randomized Adaptive Search Procedure (GRASP) was focused on the local search ability of basic genetic algorithm to generate the initial solution. The genetic algorithm was used to find the best solution from the initial solutions in the end. The calculation result shows that this improved genetic algorithm can solve the vehicle routing problem better than the basic one and reduce logistics costing effectively.