An adaptive controller was proposed for tracking control of nonholonomic mobile robots, that based on the utilization of back stepping and RBF neural networks. Firstly, an auxiliary velocity controller wss designed to make the output tracking error as small as possible. Then, an adaptive neural network controller was presented to ensure the velocity tracking ability under dynamic uncertainties and disturbances. The online adaptive laws of the RBF network were derived in the Lyapunov sense so that the stability of the system can be guaranteed. Finally, simulation results on a wheeled mobile robot are provided to show the effectiveness of the proposed approach.