An ASTER image covering the south Chocolate Mountains area, California, U.S.A. was evaluated for gold-related lithologic mapping and alteration mineral detection. A supervised classifier was first applied to the 14-channel ASTER radiance data to map lithologies related to gold deposits. Subsequently, four alteration indices were extracted from the six SWIR channels and transformed to delineate alteration zones using a PCA transformed mineralogic indices approach. Finally, a subpixel unmixing algorithm, the constrained energy minimization (CEM) technique was used to detect significant alteration minerals using the ASTER VNIR and SWIR surface reflectance data and reference spectra from the ASTER spectral library. The classification results show that the ASTER data were capable of mapping flood basalt, quartz-biotite gneiss, muscovite schist, granitic, volcanic, and metasedimentary rock units. The ASTER-derived rock units show excellent correlation with those on the reference geologic map. The overall classification accuracy is 82% and the Kappa coefficient 0.76. A group of gneisses, locally the most favorable host rocks of gold deposits, were mapped with the ASTER data with a Producer's accuracy of 86%, and more importantly were also mapped in some areas that were not shown on the field geologic map. Four alteration minerals: alunite, kaolinite, muscovite and montmorillonite were detected by subpixel unmixing analysis of the ASTER reflectance data. This study compared different methods for extracting mineralogic information from ASTER data, compared the remotely derived maps to the mapped field geology, and used the ASTER data to map minerals and lithologies related to gold exploration.