In this paper, a robust consensus tracking controller for multiple marine vessels based on dynamic surface control (DSC) and neural network (NN) is proposed. The communication topology, which is often limited due to the topography and external disturbances, is of significance for coordination of the multiple marine vessels. The influence of communication topologies, which is neglected in many control designs for coordination of multiple marine vessels, is studied in this paper. First, an independent controller is designed to force the leader to follow a predefined path. The robust distributed controller is then designed for the multiple marine vessels only using the information of itself and its neighbors. The unknown dynamics of the group of marine vessels, which are hard to be determined, are treated as one integrated nonlinear uncertainty, and the NN is employed to deal with it. In order to avoid the problem of "explosion of complexity" which is inherent in traditional backstepping technique, the dynamic surface control (DSC) is employed, in this way, the computational burden is reduced. All the signals of multiple marine vessels are proved to be uniformly bounded. The simulation results are provided to illustrate the effectiveness of the proposed control design.