This study describes how protein release from polymer matrices correlate with simple measurements on the intrinsic viscosity of the polymer solutions used for casting the matrices and calculations of the solubility parameters of polymers and solvents used. Matrices of poly(dl-lactide-co-glycolide) (PLGA) were cast with bovine serum albumin (BSA) as a model drug using different solvents (acetone, dichloromethane, ethanol and water). The amount of released protein from the different matrices was correlated with the Hildebrand and Hansen solubility parameters of the solvents, and the intrinsic viscosity of the polymer solutions. Matrix microstructure was investigated by transmission and scanning electron microscopy (TEM and SEM). Polycaprolactone (PCL) matrices were used in a similar way to support the results for PLGA matrices. The maximum amount of BSA released and the release profile from PLGA matrices varied depending on the solvent used for casting. The maximum amount of released BSA decreased with higher intrinsic viscosity, and increased with solubility parameter difference between the solvent and polymer used. The solvent used also had an effect on the matrix microstructure as determined by TEM and SEM. Similar results were obtained for the PCL polymer systems. The smaller the difference in the solubility parameter between the polymer and the solvent used for casting a polymer matrix, the lower will be the maximum protein release. This is because of the presence of smaller pore sizes in the cast matrix if a solvent with a solubility parameter close to the one of the polymer is used. Likewise, the intrinsic viscosity of the polymer solution increases as solubility parameter differences decrease, thus, simple measurements of intrinsic viscosity and solubility parameter difference, allow the prediction of protein release profiles.