Preisach model has enjoyed extensive applications in describing the hysteresis phenomena. An important open question in the analysis of hysteresis using Preisach models is the determination of the model parameters and is referred as the identification problem. However, no general mathematical methods appear to be available for identification and customized identification algorithms must be developed for each specific area of applications. In order to describe physical systems more closely, it becomes increasingly difficult to derive the Preisach function and its associated parameters from the experimental results as the complexity of hysteresis models increases. This paper presents a new approach - a wavelet identification of Preisach model. Since a wavelet can generate a basis for all functions with finite energy, in this application the system output and the Preisach functions are represented by using wavelet approximation. The coefficients of the wavelet approximation expansion can be obtained by a series of experimental data. The modeling and prediction of hysteresis can be done by manipulating the wavelet series. Some comparison of experimental and computational results are also presented in this paper.