In this paper, a new method is proposed for minimizing topographic effects on RADARSAT-1 synthetic aperture radar (SAR) images in a mountainous region. Vegetation classes in terrain-affected areas (layover and foreshortening) and the human settlement class in non-affected areas on SAR images often appear with almost the same brightness and cannot be separated. As such, SAR application to land use/land cover mapping in mountainous areas is hindered. The suggested procedures include terrain effect simulation, normalization, and slope correction for SAR images. The possibility of using RADARSAT-1 images to detect human settlements in the mountainous Three Gorges Area, China, is investigated. The performance of various radiometric correction methods are compared and evaluated. In this case study, the separability between vegetation in terrain-affected areas and human settlements was successfully improved using the proposed method. Among all the images corrected using different methods, the one corrected by the proposed method for topographic correction produced the best result for mapping human settlements and other land use and land cover classes. The producer accuracy of human settlement mapping from the SAR image increased from 16.1% before correction to 46.4% after correction, the user accuracy increased from 20.9% to 39.4%, and kappa increased from 0.13 to 0.33.