Rock material heterogeneity is one of most important intrinsic properties of geological materials. The need to quantify and model this heterogeneity is a key point for the prediction of the rock fragmentation behaviour through simulation. In this thesis both microstructure and mechanical heterogeneity are analysed. In both characterizations, statistical modelling is used as the main tool. This thesis is organized in two parts. The first part consists in overview of the approach used to quantify, model and investigate rock material heterogeneity. Moreover, a literature reviews about the influence the microstructure properties in the strength and deformability behaviour of granites is presented. Shortly, the main conclusions obtained in this thesis are: (i) after testing several statistical models it is concluded that Weibull distribution function is the most appropriate one for modelling the grain size and grain shape of the microstructure. The parameters for this modelling are given. Concerning grain orientation none of the models is able to model it as they to show a trend to anisotropy while the rock is isotropic; (ii) it is proved that even when a small number of mechanical tests are available the mechanical heterogeneity can be estimated in a reliable way. The confidence intervals for the homogeneity index of granites for unconfined compression, Young's modulus and Poisson's ratio are given. The second part of this thesis consists in four appended papers listed bellow. The first two papers (A and B) deal with the modelling of the microstructure and mechanical heterogeneity using the weakest link theory. The paper C concerns the analysis of the correlation between the microstructure and mechanical properties of aggregates.