The utility of the recently developed nonrandom two-liquid segment activity coefficient (NRTL-SAC) model has been reported for solvent selection in support of industrial crystallization process design. In this paper, we present a recent successful application with NRTL-SAC to screen solvents for a crystallization medium with the goal to maximize API solubility and to minimize solvent usage. The NRTL-SAC model parameters for the molecule in development are first identified from a minimal set of solubility experiments in selected solvents. We then perform numerous in silico virtual experiments to explore the solubility behavior of the molecule in other pure solvents and mixed solvents. The modeling results suggested optimal solvent systems for the crystallization medium which are validated in physical laboratories and chosen for process scale-up. This study demonstrated the effectiveness of the NRTL-SAC model and supports its use as a tool in drug development.