Proton exchange membrane (PEM) fuel cells operated with hydrogen and air offer promising alternative to conventional fossil fuel sources for transport and stationary applications because of its high efficiency, low-temperature operation, high power density, fast start-up and potable power for mobile application. Power levels derivable from this class of fuel cell depend on the operating parameters. In this study, a three-dimensional numerical optimisation of the effect of operating and design parameters of PEM fuel cell performance was developed. The model computational domain includes an anode flow channel, membrane electrode assembly and a cathode flow channel. The continuity, momentum, energy and species conservation equations describing the flow and species transport of the gas mixture in the coupled gas channels and the electrodes were numerically solved using a computational fluid dynamics code. The effects of several key parameters, including channel geometries (width and depth), flow orientation and gas diffusion layer (GDL) porosity on performance and species distribution in a typical fuel cell system have been studied. Numerical results of the effect of flow rate and GDL porosity on the flow channel optimal configurations for PEM fuel cell are reported. Simulations were carried out ranging from 0.6 to 1.6mm for channel width, 0.5 to 3.0mm for channel depth and 0.1 to 0.7 for the GDL porosity. Results were evaluated at 0.3V operating cell voltage of the PEM fuel cell. The optimisation results show that the optimum dimension values for channel depth and channel width are 2.0 and 1.2mm, respectively. In addition, the results indicate that effective design of fuel gas channel in combination with the reactant species flow rate and GDL porosity enhances the performance of the fuel cell. The numerical results computed agree well with experimental data in the literature. Consequently, the results obtained provide useful information for improving the design of fuel cells. Copyright (c) 2011 John Wiley & Sons, Ltd.