Mineral Resources classification by confidence levels of tonnages and grades is fundamental for Mineral Resources disclosure and to assess the maturity and risk associated with a mineral deposit. While geometric methods are transparent and easy to understand, they do not directly capture local variations of uncertainty. Geostatistical methods primarily address the quality of grade estimation. The Scorecard method aims to consider various parameters involved in Mineral Resources Classification. However, some aspects of this methodology, such as the scoring system for defining uncertainty classes and the weights as signed in the final scoring, can be considered subjective. The goal of this work is to present a methodology for Mineral Resources Classification that considers various parameters. The methodology is a hybrid approach that combines traditional geometric classification methods with risk map. The risk map is composed of sources: (i) geological uncertainty, (ii) data base quality, and (iii) grade estimation quality. To avoid subjectivity in combining uncertainties, the risk map is built by K-means clustering methodology. The risk map, divided into high, medium, and low risks. The final classification is hierarchical combination of geometric classification with the risk map. The application of subsequent machine learning methodology is used to eliminate the “salt and pepper” and/or “spotted dog” effects. The developed methodology was applied in a case study on a stratigraphic deposit and showed good results. It was able to consider traditional geometric classification with others important factors. The methodology showed to be very efficient, eliminating subjectivity in determining boundaries for classifying uncertainties and making the classification methodology reproducible.