Mineral liberation models of multi-mineral particulate materials are described by functions of continuous or discrete random variables such as the size and the grade of their particles. In the present work that describes a binary system both size and grade are treated as discrete variables and particles are classified into size and grade classes accordingly. Initially the present work develops a model to predict the distribution of particles of the same size class into the different grade classes. This distribution gives the liberation status of each size class and if combined with the size distribution of the particulate material it produces the total liberation spectrum. Initially the process defines an equivalent size, below which liberation begins, called "critical liberation size". It then describes a sequential size reduction pattern used only in order to relate the size class of the child particles to the probability of falling into any available grade class. This process defines the liberation model. The size reduction pattern accepted dictates by itself the size ratio of the equivalent size classes used. However this pattern is not used for the prediction of the volume-size (or mass-size) distribution of the final particulate material. It is believed that any size reduction model using selection and breakage functions independent of the particle grade in combination with the proposed liberation spectrum of the final product. (c) 2007 Elsevier Ltd. All rights reserved.