The main objective of this study is to develop the synthesis and optimization of reactor-separation network (RSN) models that can be simultaneously solved within Non-Linear Programming (NLP) for a PEM fuel cell system. The objective function for optimization was defined to minimize the overall cost and CO, production. Five alternatives were synthesized to determine the best flow chart for the system based on cost and the output concentration of carbon monoxide. A Polymer Electrolyte Membrane Fuel Cell (PEMFC) system was taken as the case study. The results indicated that the optimum specific cost of a PEMFC stack was found to be in the region of US$ 500-700kW(-1), while the specific manufacturing cost and the specific investment cost were calculated at the range of US$ 1000-1500 and 2500-3000 kW(-1), respectively. Furthermore, the infrastructure investment cost was determined to be in the range of US$ 10-30 billion with the specific cost for one unit in the range of US$ 2000-4000 kW(-1). The results obtained are comparable with other studies. (c) 2005 Elsevier B.V. All rights reserved.