This paper presents a new approach to model liberation for communication. The liberation distribution is characterised by a dispersion rate function, which is related to the texture of the ore. Once this function is determined the dispersion model predicts the liberation of the mill product when the feed or operating conditions changes. This model is validated using computer-simulated data and the dolomite-sphalerite data made available by Schneider (1995). The results show that the dispersion model is capable of describing and predicting the liberation of particles as size changes when breakage is nan-preferential.