A Landsat Thematic Mapper (TM) natural colour composite allowed the discrimination of areas of hydrothermally altered materials, even where vegetation (mainly herbaceous plants) covered portions of the terrain. Field spectra data showed that broad iron-oxide absorption features in TM 1 and TM 2 bands enabled the spectral discrimination between areas of hydrothermally altered materials and areas of soils derived from biotite-granites. In order to improve the definition of the target areas, the TM images were merged with a digitized aerial photograph through IHS technique. The resulting high resolution hybrid images were segmented using a region growing method, which generated images partitioned into a number of homogeneous regions. The segmented images were classified using an unsupervised clustering region classifier algorithm. The result, compared with field observations, demonstrated that the method eliminated the subjectivity of the visual image interpretation and increased the accuracy in the delineation of the hydrothermally altered areas.