A novel direct adaptive NN dynamic surface control approach is proposed for a class of typical strict-feedback nonlinear systems. The problem of explosion of terms in traditional backstepping design is eliminated by utilizing dynamic surface control. NNs are used to directly approximate the desired control input signals instead of the unknown nonlinearities in systems. The Minimax norm of all NN weight vetor is defined as updated parameter, only one parameter is needed to be estimated on-line for an n-th order strict-feedback nonlinear system. Therefore, the computation burden is significantly reduced and the possible controller singularity problem in feedback linearization is completely avoided without any additional effort. It is proved that the developed method can guarantee semi-global stability of the close-loop system. Simulation results demonstrate the effectiveness of the proposed approach.