The traditional trading of iron ores is based on chemical specifications and particle size distribution. However, recent characterisation studies, that bring additional information like mineralogical composition and microstructural (textural) aspects, have become important. They contribute to determination of the iron ores downstream beneficiation operations and subsequent steelmaking processing, allowing improvements on both new and existing processes._x000D_
Despite progress in commercial scanning electron microscope/energy dispersive spectrometer (SEM/EDS) automatic instruments in the last decade, these systems are not capable of performing the identification and discrimination of major iron ore minerals (haematite and magnetite) and the consequent description of its microstructure. On the other hand, reflected light microscopy (RLM) can easily distinguish the iron oxides by their reflectancies, but it cannot discriminate quartz and epoxy resin, which present similar colour on images._x000D_
The present work proposes an innovative method to perform a quantitative mineralogy characterisation of iron ore based on RLM-SEM co-site microscopy._x000D_
This technique combines images obtained by RLM and SEM through an automatic registration procedure that accurately aligns the two kinds of images of each field. In fact, it creates multi-dimensional images, which are then analysed using image processing and pattern recognition techniques._x000D_
The case study is an itabiritic iron ore from Quadriltero Ferrfero (Brazil) which contains mainly quartz, goethite, magnetite and haematite. The discrimination of phases that are not distinguishable with either RLM (epoxy resin and quartz) or SEM (haematite and magnetite) can be performed through this multimodal approach, allowing the subsequent mineralogical quantification._x000D_
The mineralogical quantification computed by image analysis was consistent with independently obtained results based on the Rietveld technique.