This paper is concerned with the problem of adaptive neural tracking control for a class of uncertain pure-feedback nonlinear systems. Using the implicit function theorem and back-stepping technique, a practical robust adaptive neural control scheme is proposed to guarantee that the tracking error c