Endocytosed molecules are sorted in endosomes to different cellular destinations (e.g., to lysosomes or to the plasma membrane). Diverse endosomal sorting results have been reported for different ligands and receptors in a variety of cell types, but the general principles governing these sorting outcomes are not well understood. For example, we observed a wide range of sorting outcomes with the epidermal growth factor (EGF)/receptor system in fibroblasts using several members of the EGF family and site-directed ligand and receptor mutants. In this article we describe a mechanistic mathematical model of endosomal sorting based on the hypothesis that receptors may be selectively retained by the endosomal sorting apparatus and that this process may be modulated by receptor occupancy. Our results show that this single mechanism can account for the wide variety of observed sorting outcomes. By providing a conceptual framework for understanding endosomal sorting, this model not only helps interpret our experimental results for the EGF/receptor system, but also provides some insight into the principles governing sorting. For example, the model predicts that the influence of selective endosomal retention of receptor/ligand complexes is seen in deviations of ligand sorting outcomes from pure fluid phase sorting behavior. Furthermore, the model suggests that selective endosomal retention of complexes within endosomes gives rise to three sorting regimes characterized by distinguishable qualitative trends in the dependence of ligand sorting fractions on intracellular ligand concentrations.