Composition modeling using constrained homologous series permits the derivation of the detailed composition of complex mixtures starting from a limited set of mixture bulk properties. By imposing various constraints, the number of unknowns is drastically reduced. Probability density functions are imposed on both the carbon number distribution in each homologous series of components and on the structural attribute distributions. Validation, based on detailed compositional information, shows that the use of gamma distributions to constrain the mixture composition results in an adequate approximation of the experimentally measured composition. The latter was obtained for two middle distillates and a heavy gas oil using advanced analytical techniques.