A strategy to infer solubilities from the combination of experiment and all-atom simulations is presented. From a single experimental estimate, the solubility of a substrate can be predicted in various environments from the related free energies of solvation. In the case of quercetin, the methodology was shown to reproduce the experimental solubilities in chloroform, water, acetonitrile, acetone, and tert-amyl alcohol within 0.5 log unit. The reliability of the estimates is markedly correlated to the accuracy of the experimental measure and to both the accuracy and precision of the computed free energies of solvation.