A novel adaptive radial basis function(RBF) neural network sliding mode strat egy is developed to con trol Lorenz chaos with parametric uncertainties and external disturbances. Based on the controllable canonical form of system state error at its unstable equili brium, a sliding surface is defined as the only input to the RBF controller. Onl y seven RBFs are required for the controller and their weights ar e trained on-line based on the sliding surface approaching condition. The simula tio n results show that this method is feasible and effective, and the robustness to parametric uncertainties and external disturbance is provided.