The processing and formulation of pharmaceutical, medical and agrochemical products often requires the solubility of an active substance in liquids. Being able to provide fast and reasonably reliable property information, either from measurement or prediction, can greatly enhance the competitiveness of formulations. However, due to the complexity of many active ingredients and solvents, reliable measurements are often unavailable while existing property prediction models cannot provide sufficiently accurate predictions. Therefore most design work in this field is done by trial-and-error and empirical descriptions of limited experimental results. The result is that industrial process groups spend as much as 30% of their time on solubility-related problems. To alleviate the inefficiency and errors of this situation, we have taken a simplifying approach to determining changes in the solubility of sparingly soluble complex chemicals as the solvent type and composition are varied.