Floods cannot be prevented but accurate information about them can reduce its severity to large extent. Remote sensing technology in general and SAR images in particular are known to be efficient and cost-effective when large areas affected by floods are to be mapped and analyzed. The main objective of the present work was to make a comparative analysis of different classification techniques for flooded area extraction on RADARSAT images. Assessment of accuracy and validation by appropriate method along with role of ancillary data for comparison of results were other issues discussed.