Genetic engineering has been used for the fusion of peptides, with different length and composition, on a protein to study the effect on partitioning in aqueous two-phase systems containing thermoseparating polymers. Peptides containing 2-6 tryptophan residues or tryptophan plus 1-3 lysine or aspartate residues, were fused near the C-terminus of the recombinant protein ZZT0, where Z is a synthetic IgG-binding domain derived from domain B in staphylococcal protein A. The partitioning behavior of the peptides and fusion proteins were studied in an aqueous two-phase system composed of dextran and the thermoseparating ethylene oxide-propylene oxide random copolymer, EO30PO70. The zwitterionic compound p-alanine was used to reduce the charge-dependent salt effects on partitioning, and to evaluate the contribution to the partition coefficient from the amino acid residues, Trp, Lys, and Asp, respectively. Trp was found to direct the fusion proteins to the EO-PO copolymer phase, while Asp and Lys directed them to the dextran phase. The effect of sodium perchlorate and triethylammonium phosphate on the partitioning of the fusion proteins was also studied. Salt effects were directly proportional to the net charge of the fusion proteins. Sodium perchlorate was found to be 3.5 times more effective in directing positively charged proteins to the EO-PO copolymer phase compared to the effect of triethyl ammonium phosphate on negatively charged proteins. An empirical correlation has been tested where the fusion protein partitioning is a result of independent contributions from unmodified protein, fused peptide, and salt effects. A good agreement with experimental data was obtained which indicates the possibility, by independent measurements of partitioning of target protein and fusion peptide, to approximately predict the fusion protein partitioning.