In many cities, large trucks can't be used to delivery in the urban district, but in the suburbs all kinds of vehicles can be used. To solve the problem, this paper establishes the optimization model and designs a solving method using a genetic algorithm (GA), taking Zigong tobacco distribution center for example. Comparing with the data calculated by empirical distribution route, the results show that the total distance declines 155.47 km, the number of different type vehicles reduces 1-2 and the average load rate increased by 68.3% to 85.1%.