Scanning electron microscope-based automated image analysis (SEM-AIA) is a technique of characterizing particles for size and shape from SEM images and for elemental composition from energy-dispersive X-ray spectra, and because it is automated it can provide data for a statistically significant number of particles. It was used to characterize the association of mineral matter with organic coal components in order to predict cleanability for both density-based and surface-based cleaning processes. Selected samples of coals from Pittsburgh No. 8, Upper Freeport, and Illinois No. 6 seams, ground by two different research groups to a nominal 200-mesh particle size, were characterized using this technique. SEM-AIA results indicated significant differences in the overall particle size distributions between the Pittsburgh No. 8 coal samples from the two groups, which were reflected in the association of coal and mineral matter. Coal and mineral phases within composite coal-mineral particles were characterized in cross section for phase size and identity. The amount of phases present within each composite particle and the combined amount of mineral phases present on the exposed perimeter were also measured as indicators of association with regard to density- and surface-based cleaning processes, respectively. For the three coals ground at the same place to comparable sizes, the Upper Freeport appeared to be the most cleanable by density-based processes, followed by the Illinois No. 6 and then the Pittsburgh No. 8 coal. For the same coals, the SEM-AIA results also predicted significantly less cleanability for surface-based processes; at high levels of coal recovery and mineral reduction, 10-20% less mineral reduction was predicted at the same level of coal recovery. These differences are apart from any surface chemistry considerations and involve only the association, i.e., the distribution of coal and mineral matter within the particles. The coals exhibited the same relative ranking of cleanability by density- or surface-based processes.