The binding of solvent molecules to a protein surface was modeled by molecular dynamics simulations of of Candida antarctica (C. antarctica) lipase B in binary mixtures of water, methanol, and toluene. Two models were analyzed: a competitive Langmuir model which assumes identical solvent binding sites with a different affinity toward water (K-Wat), methanol (K-Met), and toluene (K-Tol) and a competitive Langmuir model with an additional interaction between free water and already bound water (K-WatWat). The numbers of protein-bound molecules of both components of a binary mixture were determined for different compositions as a function of their thermodynamic activities in the bulk phase, and the binding constants were simultaneously fitted to the six binding curves (two components of three different mixtures). For both Langmuir models, the values of K-Wat, K-Met, and K-Tol were highly correlated. The highest binding affinity was found for methanol, which was almost 4-fold higher than the binding affinities of water and toluene (K-Met >> K-Wat approximate to K-Tol). Binding of water was dominated by the water-water interaction (K-WatWat). Even for the three protein surface patches of highest water affinity, the binding affinity of methanol was 2-fold higher than water and 8-fold higher than toluene (K-Met > K-Wat > K-Tol). The Langmuir model provides insights into the protein destabilizing mechanism of methanol which has a high binding affinity toward the protein surface. Thus, destabilizing solvents compete with intraprotein interactions and disrupt the tertiary structure. In contrast, benign solvents such as water or toluene have a low affinity toward the protein surface. Water is a special solvent: only few water molecules bind directly to the protein; most water molecules bind to already bound water molecules thus forming water patches. A quantitative mechanistic model of protein-solvent interactions that includes competition and miscibility of the components contributes a robust basis for solvent and protein engineering.