We have quantitatively characterized nearest- and near-neighbor distances and local-area fractions of second-phase populations in a matrix using automatic image analysis and Dirichlet cell tessellation procedures. These procedures were applied to distributions of sulfide inclusions and carbides in six steels and to a composite of graphite fibers in aluminum. In addition, computer-generated random particle distributions were obtained for the different steels and the composite using a distribution of particle sizes with the same means and standard deviations as the actual distributions. Comparison of the actual second-phase particle distributions with the generated distributions showed that the sulfide distributions consisted of clusters superimposed on a random spatial distribution, while the carbide distributions and the graphite-fiber distributions are close to random.