Improving the performance of polymer-electrolyte membrane (PEM) fuel cells depends on the optimization of catalyst layer composition and structure for large active surfaces. Modelling studies provide a valuable tool for investigating the effects of catalyst layer composition and structure on the electrochemical and physical phenomena occurring in PEM fuel cells. Previous modelling studies have shown that the distribution of electrochemical reactions in catalyst layers is highly dependent on the complex interaction of activation and ohmic effects as well as contributions from transport limitations and variations in local and overall current densities. In this paper, three-dimensional, multicomponent and multiphase transport computations are performed using a computational fluid dynamics (CFD) code (FLUENT (TM)) with a new PEM fuel cell module, which has been further improved by taking into account the detailed composition and structure of the catalyst layers using the multiple thin-film agglomerate model. The detailed modelling of reactions in the catalyst layers is used to determine methods of improving the effectiveness of catalyst layers for a given platinum loading. First, available data on catalyst layer composition and structure are used in CFD computations to predict reaction rate distributions. Based on these results, spatial variations in catalyst loading are then implemented in CFD computations for the same overall catalyst loading to investigate possible performance gains. It is found that grading catalyst loading towards the membrane in the anode and the gas channel inlet in the cathode provides the most beneficial effects on the fuel cell performance. Thus the results suggest that significant savings in cost can be attained by reducing the platinum loading in underutilized regions of the catalyst layers, while at the same time improving the performance. Copyright (C) 2009 John Wiley & Sons, Ltd.