Previously Vilker et al. (J. Colloid Interface Sci. 79(2), (1981)) reported the osmotic pressure of concentrated bovine serum albumin (BSA) up to 475 g/L in 0.15 M sodium chloride at pH 4.5, 5.4, and 7.4. The authors used a semiempirical model based on Donnan theory to predict the osmotic pressure with good agreement. However, the formal application of a three-term virial expansion with the coefficients determined from the potential energy of interaction between BSA molecules resulted in poor agreement with their data. In this study, modeling of the osmotic pressure was reexamined using a free-solvent model that considered average solute-solvent and microion-solute interactions in a mole fraction concentration variable. The resulting fits were excellent for all three pH. The model is designed with no fitted parameters; however, the model results were highly sensitive to the selected hydration and microion binding. Therefore the hydration was further regressed from its initial estimate of I g H2O/g BSA (based on water-O-17 magnetic resonance studies of other globular proteins) to minimize the least-squares error between the predicted values and data. The resulting average hydration was determined to be 1.14 +/- 0.03 g H2O/g BSA for all pH values. However, the standard error in hydration for each pH was no greater than +/-0.0063 g H2O/g BSA. These results demonstrate that solvent-solute interaction and the concentration variable may be critical factors when evaluating osmotic pressure data of concentrated protein solutions.