A possible output of the European project ReSeDA is to use the very complete available data base covering one year of the growing season for assessing the interest of multi-angular time series of radar images (Baret & al, 1999). The aim is to prepare an optimal use of A-SAR data for the purpose of regional agricultural statistics on a site representative of semi-irrigated areas. Four sets of time series of SAR data have been compared in the perspective of operational assessment: 3 dates of ERS2, 3 dates of Radarsat, 6 dates of bi-angular ERS and Radarsat, and 9 dates of ERS. Two filters and a software of segmentation of images (PAPRI from CIRAD) have been successively applied over the 4 sets. The results of classification are compared with the reference plots of the test site Alpilles for 3 major annual crops (wheat, maize and sunflower). Then the robustness of the best discrimination is tested at a regional scale through 12 external segments of 700x700 m, according to the well known sampling system of the european MARS project. The pre-segmentation of radar images and the bi-angular series of SAR improve the discrimination annual crops for assessing their areas at a regional scale. It gives robust guidelines about the interest of SAR data for the regional crop distribution in case of cloudy conditions.