The general vehicle routing problem has been studied by many researchers such as Christofides, et al. and Laporte, et al., but only limited effort has been devoted to developing the optimal algorithms. The purpose of this paper is to develop a branch and bound algorithm which determines the optimal vechicle routes and the optimal number of vehicles concurrenetly for the asymmetrical vehicle routing problem. In order to enhance the efficiency, this algorithm emphasizes the followings ; First, an efficient primal-dual approach is developed to solve subproblems which are called the specialized transportation problem, formed by relaxing the illegal subtour constraints from the vehicle routing problem, second, an improved branching scheme is developed to reduce the number of candidate subproblems by adequate utilization of vehicle capacity restrictions.