The goal of this paper was two-folded. Firstly, based on the available literature, descriptive models were compiled for generic stratabound, sedimentaryhosted Zn-Pb deposits, and optimum spectral bandwidths covered by contemporary multispectral sensors were suggested for detection of their primary geologic features. Secondly, the detection model was tested in the Salobro Zn-Pb deposit (N-NE portion of the Minas Gerais State), considering GEOSCAN data that consist of 24 spectral bands at 5m spectral resolution. Before the conceptual detection model was tested in the Salobro deposit, a geologic surveying followed by petrographic and spectral analysis were accomplished in order to tune the model to a list of local geologic observational phenomena at the surface (e.g., favourable host rocks, alteration patterns, structural controls) that might lend themselves to remote sensing investigation. The list of observational phenomena were then filtered by a set o physical environmental constraints (climate, vegetation and soil cover) to produce a new set of landscape attributes (detectable phenomena) that stood a reasonable chance of being detected and exploited in this particular study area. Dense vegetation (even in dry seasons) and soil (either in situ or transported) cover most of rocks throughout the area of the deposit, which limits considerably the observational features, screening the detectable features to a few. Among the main detectable features are the Zn-Pb ore zone (ferruginous metachert) and banded iron formations closely associated to it. Key spectral bandwidths to detecting these two sets of rocks and that are simultaneously available within GEOSCAN data, comprise : (i) 400-950nm– covering the visible and near infrared region of the spectrum, for mapping iron oxides and hydroxides; and (ii) 8500-12500nm– covering the thermal region, for mapping silica-rich rocks. A reasoned thematic mapping approach, favoured in this study, tailored image processing of GEOSCAN data to the specific attributes of interest, focusing on the detectable features yielded from the model. In this view, image processing was split in two steps: (i) a basic toolkit for image processing, including colour composite images, band ratios and principal component transformations, were applied to the data aiming to discriminate between the key rocks of the deposit; (ii) spectral classifiers (SAM and SFF) were then employed to identify such rocks based on spectral libraries. As predicted by the detection model, both sets of techniques, particularly the ordinary ones (e.g., band 20 - 9170nm ± 530nm; band 14 - 2176nm ± 44nm and band 6 - 740nm ±23nm, in RGB, merged with the digital elevation model of the area), were able to successfully map the surface expression of the ore zone and the banded iron formations within the deposit. However, most of the other geologic features associated to the deposit was Masked by vegetation and soil cover. This paper has demonstrated that theoretical exploration models based on remote sensing data can sufficiently support the indirect targeting of base metal deposits. However, the physical environment at the surface, as well as the choice of remote sensing data, may constrain the suitability of the model for a particular scale. Using the Salobro deposit as a control, this work showed that the application of a specific detection model, coupled with the moderately high spatial and spectral resolution of GEOSCAN data, was able to frame the ore zone accurately, an achievement that has been rarely repeated in tropical terrains.