Model based solvent selection can help to point out green alternatives to environmentally unfriendly solvents by examining a much wider range of solvent space than is possible with even high throughput measurements. Use of a model for solvent selection can also help to cut down the number of experiments required for solvent selection so that less waste is generated. The NRTL SAC (Non-Random Two Liquid Segment Activity Coefficient) model is a parameter based method for estimating solubilities in different solvents which can be used for solvent selection. It has been widely applied at Bristol Myers Squibb in over 100 compounds for this purpose. Typically, five to ten solubility measurements are used as input to the model and 0.25 to 0.5 grams of material is required for these measurements. In order to provide solubility predictions earlier in development using less material and further cut down on waste generation, a model for predicting solubility data using chromatographic retention times as model input has been developed based on the NRTL-SAC solubility model. Using the model, NRTL-SAC model parameters were established for twelve chromatographic stationary phases from retention times of molecules with known NRTL-SAC model parameters. Based on these model parameters for the stationary phases, retention times were used to find the previously unknown model parameters for a set of test molecules which were then used to predict their solubilities. These predictions as well as case studies in model-based solvent selection will be presented.