In this paper, adaptive neural network (NN) control is presented for a class of non-affine nonlinear systems with unknown control gain. Using the Mean Value Theorem and backstepping method, we propose a constructive approach for adaptive NN control design. The Nussbaum function is used to relax the requirement on control direction. The semi-global uniformly ultimately boundedness (SGUUB) of the closed-loop system is achieved. In addition, the simulation study results are given to demonstrate the effectiveness of the proposed control.