Solubility models for four protein-salt systems have been developed with the aid of an artificial neural network technique. The solubility of proteins in salt solutions is a complex phenomenon dependent on the type of protein, pH, temperature, concentration, and type of salt. In these models, the solubility has been correlated as a function of temperature, pH, and salt concentrations. The four systems are carboxyhaemoglobin in potassium phosphate solutions, ovalbumin in ammonium sulfate solutions, glucose isomerase in ammonium sulfate solutions, and concanavalin A in ammonium sulfate solutions. The models predicted the solubilities with an average quadratic error ranging from 0.00025 to 0.002. The model predictions were then analyzed to study the effect of pH, temperature, and salt concentrations. The predictions were found to be qualitatively in agreement with reports from the literature.