This paper focuses on the analysis of the statistical segmentation methods of SAR images. We present here a 2 years work of the national French research group on the classification of SAR images. This group gathers about 5 remote sensing research teams and 5 image processing laboratories. The collaboration between the two communities allowed to get an overview of the segmentation methods for SAR images and to point out the advantages of modelling, estimation methods or strategies of classification used in the different algorithms. This paper presents also the methodology of quantitative evaluation of the algorithms performances and the analysis of the remote sensing experts.