Ultrafiltration process was used for olive oil mill wastewater treatment in this work. Filtration experiments were performed with raw wastewater and no pretreatment was applied. Effects of major operating variables such as the transmembrane pressure, feed flow rate and operation time were investigated. A Box-Wilson statistical experiment design method was used by considering the transmembrane pressure (1-3 bar), recycling flow rate (100-200 L/h) and operation time (30-120 min) variables while COD removal and permeate flux variation were the objective functions. The results were correlated by a response function and the coefficients were determined by regression analysis. COD removal efficiency and permeate flux variation determined from the response functions were in good agreement with the experimental results. The optimum set of flow rate and pressure for COD removal was 100 L/h and I bar with 89.5% COD removal efficiency. On the other hand, optimum set was 200 L/h and 3 bar for permeate flux with 34.4 L/(m(2) h) flux value. Since optimum sets are just at opposite corners of operation region (pressure, flow rate), first of all objective (COD removal or permeate flux) should be determined and then optimum operation parameters can be selected. (C) 2008 Elsevier B.V. All rights reserved.