The polymer assisted ultrafiltration process combines the selectivity of the chelating agent with the filtration ability of the membrane acting in synergy. Such hybrid process (complexation-ultrafiltration) is influenced by several factors and therefore the application of experimental design for process optimization using a reduced number of experiments is of great importance. The present work deals with the investigation and optimization of cobalt ions removal from aqueous solutions by polymer enhanced ultrafiltration using experimental design and response surface methodological approach. Polyethyleneimine has been used as chelating agent for cobalt complexation and the ultrafiltration experiments were carried out in dead-end operating mode using a flat-sheet membrane made from regenerated cellulose. The aim of this part of experiments was to find optimal conditions for cobalt complexation, i.e. the influence of initial concentration of cobalt in feed solution, polymer/metal ratio and pH of feed solution, on the rejection efficiency and binding capacity of the polymer. in this respect, the central compositional design has been used for planning the experiments and for construction of second-order response surface models applicable for predictions. The analysis of variance has been employed for statistical validation of regression models. The optimum conditions for maximum rejection efficiency of 96.65% has been figured out experimentally by gradient method and was found to be as follows: [Co2+](0) = 65 mg/L, polymer/metal ratio = 5.88 and pH 6.84. (C) 2009 Elsevier B.V. All rights reserved.