A correlation between the osmotic second virial coefficient and the solubil
ity of proteins is derived from classical thermodynamics to support an empi
rical relation previously found by Wilson and co-workers (1). The model is
based on the equality of fugacities of the protein in the equilibrium phase
s, with the details of the model depending on the standard state used. The
parameters in this model have been fitted to data for several systems, main
ly with lysozyme as the protein. The model is found to describe experimenta
l data, with variations in protein concentration, salt type and concentrati
on, temperature, and pH, bath qualitatively and quantitatively. Agreement b
etween the model and the experimental data is very good for protein solubil
ities up to 30 mg/mL. Above this value the model underpredicts the experime
ntal data, probably as a result of multibody interactions that are not incl
uded in the model here. Variations of the model parameters with protein typ
e, temperature, pH, and salt type are discussed.