Mineral processing plants rely on mechanical separation processes to separate particles containing ore minerals from those devoid of ore minerals, i.e. comprised of gangue minerals. In order to maximize the recovery of ore minerals in a given mineral separation unit, it is essential to understand the intrinsic separation behavior of particles in the feed stream. Machine learning-based models have only recently been proposed to predict particle separation behavior by utilizing their actual composition, including attributes such as actual mineralogy, geometry and liberation. The particle composition data that provides the fundament for such prediction models is currently obtained through scanning electron microscopy-based image analysis methods. However, SEM-based image analysis methods are necessarily limited to 2D; data are, therefore, prone to carry a systematic stereological bias, which can compromise the robustness of separation behavior prediction. To overcome the limitations of 2D mineral characterization, the application of X-ray computed tomography (X-ray CT) has been successfully explored for determining particle mineralogy and geometry in 3D. However, a standardized X-ray CT-based method has yet to be developed. Recently, a new method called Mounted Single Particle Mineralogical Analyses (MSPaCMAn) has been introduced, which utilizes the grey values and geometry of individual particles for 3D characterization. MSPaCMAn aims to be the first standardized and semi-automated method for quantitative 3D particle characterization. In this contribution, we will present the results of applying MSPaCMAn to particulate samples of iron ores containing magnetite, hematite, apatite and silicate gangue. Furthermore, the results obtained through CT are compared to data obtained by quantitative X-ray powder diffraction (XRD) and SEM-based image analysis.