Abstract: The existing adaptive structure are not suitable to update the parameters for Hammerstein predistorter, especially for the linear subsystem. In order to solve the problem, an improved adaptive structure based on indirect learning structure is proposed in this paper. Efficient least-square algorithm can be used to update parameters of the linear subsystem in a Hammerstein predistorter by this adaptive structure which can provide the error of the linear subsystem. Therefore, the linearization efficiency of the whole predistorter is improved. The experimental results show that the Hammerstein predistorter by using the proposed adaptive structure can compensate the nonlinear distortion of PA with memory effects efficiently.