A review is given of the present applications of random sets and integral geometry in fragmentation and liberation models of ores and minerals. Although attempts have been made in the past to apply such mathematical techniques to the comminution of solids, it is shown that it would not result in major progress. In liberation modeling, it is demonstrated that the approach can lead to fruitful results. Following a recent paper by Davy (1984), a model of liberation prediction is developed that can be calibrated for ore texture and breakage using image analyzers. The model is demonstrated for the case of a Poisson polyhedra ore texture and monodisperse fragment size. The model enables prediction of the complete composite particle distribution and thus is valuable for integration of ore breakage models with minerals separation models.