We have examined the effect of genetically engineered charge modifications on the partitioning behavior of proteins in dextran/polyethylene glycol two-phase systems containing potassium phosphate. By genetically altering a protein’s charge, the role of charge on partitioning can be assessed directly without the need to modify the phase system. The charge modifications used are of two types : charged tails of polyaspartic acid fused to beta-galactosidase and charge-change point mutations of T4 lysozyme which replace positive lysine residues with negative glutamic acids. The partition coefficient K for these proteins was related to measured interfacial potential differences Delta phi using the simple thermodynamic model, ln K-p = ln K-o + (F/RT)Z(p) Delta phi. The protein net charge Z(p) was determined using the Henderson-Hasselbalch relationship with modifications based on experimentally determined titration and isoelectric point data. it was found that when the electropartitioning term Z(p) Delta phi was varied by changing the pH, the partitioning of lysozyme was quantitatively described by the thermodynamic model. The beta-galactosidase fusions displayed qualitative agreement, and although less than predicted, the partitioning increased more than two orders of magnitude for the pH range examined. Changes in the partitioning of lysozyme due to the various mutations agreed qualitatively with the thermodynamic model, but with a smaller than expected dependence on the estimated charge differences. The beta-galactosidase fusions, on the other hand, did not display a consistent charge based trend, which is likely due either to the enzyme’s targe size and complexity or to nonelectrostatic contributions from the tails. The lack of quantitative fit with the model described above suggests that the assumptions made in developing this model are oversimplified.