Abstract: For it is very difficult and complex to solve large-scale dynamic vehicle routing problem on distribution goods, propose the multi-objective optimization Dynamic Vehicle Routing Problem with Time Windows (DVRPTW) model for distribution goods, which maximizes the number of customer serviced, minimizes customer waiting time and the total vehicle driving distance and covers dynamic information both random demand and dynamic network. Then a two stage algorithm model based on hill-climbing and genetic hybrid algorithm is designed to solving DVRPTW. At last, we do the simulation experiment with standard test data from Solomon and the result shows that this model and algorithm is quite capable of solving the dynamic vehicle routing problem on distribution goods.