Ternary phase diagrams were measured for lysozyme in ammonium sulfate solutions at pH values of 4 and 8. Lysozyme, ammonium sulfate, and water mass fractions were assayed independently by UV spectroscopy, barium chloride titration, and lyophilization respectively, with mass balances satisfied to within 1%. Protein crystals, flocs, and gels were obtained in different regions of the phase diagrams, and in some cases growth of crystals from the gel phase or from the supernatant after floc removal was observed. These observations, as well as a discontinuity in protein solubility between amorphous floc precipitate and crystal phases, indicate that the crystal phase is the true equilibrium state. The ammonium sulfate was generally found to partition unequally between the supernatant and the dense phase, in disagreement with an assumption often made in protein phase equilibrium studies. The results demonstrate the potential richness of protein phase diagrams as well as the uncertainties resulting from slow equilibration.