The Vehicle Routing Problem(VRP)is a very important operation within a multiechelon warehousing and chains stores system﹒The problem of VRP is to transport aecessing stocks or pre-odering from the warehouse to the retailer with a minimum transportation cost(or minimum transportation distance)and satisfying some restrictive constraints﹒ The problem that we are dealing with is to add a new constraint which will retrict the vehicle to pickup certain merchandise for the manufacture before delivery﹒In other words,there are some precedent constraints in these cities﹒The problem is very often confronted by the dispatch of the warehouse﹒In this research, we use neural network technique to formulate the problem,then the constraint satisfactory approach also is proposed to solve the problems﹒By combining the heuristic in TSP and technique in the neural network,we will develop neural network technique which will quickly find a feasible tour(satisfactory precedence constraints)with the transportation cost does to optimal﹒The solution quality will be verified by the experimental tests and further application of this model will be discussed﹒