Abstract: A new classification method of 2-frequency pol-SAR images is proposed in this paper. This method is based on that the scattering properties vary with the changing of frequencies. The feature variation parameter and feature variation plane, which can be used for targets classification, are defined in this paper. The parameters and are used in this paper and the plane of is divided into 9 different areas standing for 9 classes. This classification method is simple both to understand and to apply. Moreover, a new unsupervised classifier is developed by combining this method with the Wishart classifier. The new unsupervised classifier is applied to the measured NSAS/JPL SIR-C data, and satisfied result is obtained. As a conclusion, this classification method is effective to pol-SAR images.