A predictive model was developed that captures quantitative structure-property relationships between the molecular structures of surfactants and co-solvents and microemulsion phase behavior. Both the optimum salinity and the optimum solubilization ratio (and thus the interfacial tension) are modeled as a function of both monovalent and divalent cations in the brines. A dataset consisting of 685 microemulsion phase behavior experiments with 24 unique crude oils, 85 surfactants (internal olefin sulfonates, alkylbenzene sulfonates, alcohol alkoxy sulfates and alcohol alkoxy carboxylates) and 18 co-solvents (alcohols and alcohol alkoxylates) was used for the model development and validation. Variations in the type of hydrophobe (carbon number, degree of branching, polydispersity, and aromaticity), number of propylene oxide groups, number of ethylene oxide groups, and the type of head group (sulfonate, benzene sulfonate, sulfate, carboxylate, hydroxyl) were studied. The oils were characterized using their equivalent alkane carbon number. The model includes the effect of soaps generated from the neutralization of acidic crude oils. The interfacial concentration of co-solvent is calculated using the pseudophase model. Previous models for optimum salinity have not included the effects of divalent cations, soap, and co-solvents among other limitations. Most importantly, the new model can be used to predict interfacial tension as well as optimum salinity and provide a better understanding of the impact of molecular structures on microemulsion phase behavior. The model is sufficiently accurate to provide a useful guide to experimental testing programs for the development of chemical formulations for enhanced oil recovery and other similar applications requiring low interfacial tension.