In tropical regions, abundant in iron-rich geological materials, caves that are genetically and geographically associated with exploitable mineral deposits may develop. These caves have speleological relevance and are environmentally and legally protected in Brazil. Thus, for better planning of exploitation and environmental licensing, it is necessary to study the genesis and development of the iron formation caves seeking to preserve them without impeding the advancement of mining. This subject is complex, rarely studied, and few are the knowledges on alternatives to predict the occurrence of these caves. This gap justifies the development of research and products capable of assisting decision-makers, planners, and competent authorities in supporting the definition of target sites for speleological prospecting in the field. In this study, the prediction of the factors involved in the development of iron formation caves was evaluated producing a map of susceptibility in a GIS environment using fuzzy logic and an analytical hierarchic process (AHP). Therefore, the variables: iron oxide ratio, slope gradient, normalized difference vegetation index (NDVI), plan curvature, profile curvature, lineament intensity, and height above the nearest drainage (HAND) were selected. These variables were obtained by processing of geospatial data from a region of the Gandarela Range (Minas Gerais, Brazil). The fuzzy logic and AHP techniques were applied, and for the validation of the results, a previously surveyed cave inventory was used. The results showed satisfactory performance of the map produced in predicting areas favorable to the occurrence of iron formation caves, presenting an area under the receiver operating characteristic (ROC) curve of approximately 0.85, which indicates a high prediction rate and validates the proposed method. Such results demonstrate that this susceptibility map was reliable and that the set of criteria and weights used were suitable for mapping areas favorable for speleological prospecting.