The method of holotomography was applied to investigate the microstructure of a metal–matrix composite (MMC) consisting of an AA6061 aluminium matrix reinforced with 20 vol.% Al₂O₃ particles. Various statistical functions characterizing the Al₂O₃ particles and the matrix were evaluated, which facilitate a better understanding of the correlation between microstructural features and the overall physical properties of the composite. The shape of the particles is approximated by an ellipsoid and their spatial distribution is described in terms of the pair-distribution function of particle centers and the two-point probability functions of the component phases. The investigated statistical functions emphasize the existence of a short-range correlation in the microstructure, which has a correlation length of about two times the average nearest-neighbor distance. To understand better the influence of the particle shape on different statistical descriptors a comparison between the real microstructure and a system of random hard-spheres is presented.