Xylose reductase (XR) from Debaryomyces hansenii was extracted by partitioning in aqueous two-phase systems (ATPS) composed of polyethylene glycol (PEG) 4000 in the presence of different salts, specifically sodium sulfate, lithium sulfate and potassium phosphate. Batch extractions were carried out under different conditions of temperature (25-45 degrees C) and tie-line length (TLL) for each system, according to a central composite design face-centered of 36 tests, and the response surface methodology was used to evaluate the results. Quadratic polynomial models were adjusted to the data to predict the behavior of four responses, namely the XR partition coefficient (K(XR)), the selectivity (S), the purification factor (PF(T)) and the activity yield (Y(T)) in the top phase. The optimal extraction conditions were found using the PEG 4000/sodium sulfate system at 45 degrees C and TLL=25.1, which ensured PF(T)=3.1 and Y(T)=131%. The ATPS proved effective for partial purification of D. hansenii xylose reductase in cell-free crude extract, and the response surface methodology revealed to be an appropriate and powerful tool to determine the best dominion of temperature and ATPS composition.