In this paper, a dynamic surface control scheme based on adaptive support vector regression(SVR) is proposed for a class of nonlinear strict-feedback systems.First of all, the kernel function of SVR is expended in a first-order Taylor series near the estimated kernel width and support vector, such that the widths and support vectors can be linearly expressed.Then, the unknown system dynamics is modeled by employing the developed SVR, and the virtual controllers and controllers are designed based on the modeling results.Meanwhile, in order to improve the modeling accuracy, prediction variables of system states and tracking errors are added in the controllers design, and the parametric adaptive laws are developed according to the prediction errors.Finally, the analysis of uniform ultimate boundedness of the system is given based on the Lyapunov theorem.Simulation results show that the proposed method can reduce the modeling error and improve the tracking accuracy effectively.