Multiobjective optimization method was applied to achieve high quality wines using membrane filtration. Different Hungarian wines were filtered by ceramic cross-flow membrane filters with changed recycle rates. The quality parameters of wines (polyphenol, protein, colloid and tartaric acid concentrations, etc.) and the permeate flux were measured as functions of membrane pore size, recycle flow rate and pretreatment of wine. The quality and operating parameters were expressed by regression functions for each wine. To calculate the multiobjective optimum the minimal loss method was used with weight factors. Case studies are shown for producing bottled wine and champagne taking into account at the same time quality and economic purposes. (C) 2003 Elsevier Science Ltd. All rights reserved.