The Tapajós Mineral Province (TMP), located in the central-south region of Amazonian Craton, stands out for its gold production for over six decades and has demonstrated huge potential for occurrences of magmatic-hydrothermal polymetallic deposits, especially epithermal and porphyry types. However, even with all the efforts made in the last few decades, including the recent publication of the province geologic map, the geologic knowledge is uncomplete, a reality that disfavor the development of the mineral exploration in the region. This situation is aggravated by the natural difficulties that this region imposes, such as access difficulties, dense forest cover and thick weathering profile.The use of indirect mapping techniques in regions with similar difficulties is expanding in the last few years. In this context, predictive mapping based on Machine Learning techniques, using different datasets, has proved to be a useful tool to characterize and delimitate geological units, recognize compositional anomalies, geological structures and more.In this contribution, aerogeophysical data is processed and interpreted to investigate the geology of the eastern region of the TMP, in the light of the geologic map.The obtained data were utilized as parameters for the production of a predictive geologic map using Self-Organizing Maps (SOM), an unsupervised Machine Learning technique, which analyses and structure the data independently, clustering the samples according to their similarities, which is recognized by the algorithm itself.The results evidence the strong contribution from the aerogeophysical data in the production of the geologic map, with strong correlation between the contacts of both maps, attesting a successful outcome from the use of SOM in this study. However, in many regions of the geologic map, defined geological units and structures are not supported by the data obtained in this study, especially in regions with few or no-data from previous field surveys.The SOM technique was effective in the creation of a predictive geologic map and appears to be more efficient that the use of geophysical data alone to map previously described geological units and to infer units in not-studied regions. It also has potential to infer areas affected by hydrothermal activity, which can be useful in the elaboration of exploratory models and defining prospective areas for exploration.