The effect of the cathode catalyst layer's structure and composition on the overall performance of a polymer electrolyte membrane fuel cell (PEMFC) is investigated numerically. The starting point of the sub-grid scale catalyst layer model is the well-known flooded agglomerate concept. The proposed model addresses the effects of ionomer (Nafion) loading, catalyst (platinum) loading, platinum/carbon ratio. agglomerate size and cathode layer thickness. The sub-grid scale model is first validated against experimental data and previously published results, and then embedded within a two-dimensional validated computational fluid dynamics code that can predict the overall performance of the fuel cell. The integrated model is then used to explore a wide range of the compositional and structural parameter space, mentioned earlier. In each case, the model is able to correctly predict the trends observed by past experimental studies. It is found that the performance trends are often different at intermediate versus high current densities-the former being governed by agglomerate-scale (or local) losses, while the latter is governed by catalyst layer thickness-scale (or global) losses. The presence of an optimal performance with varying Nafion content in the cathode is more clue to the local agglomerate-scale mass transport and conductivity losses in the polymer coating around the agglomerates than due to the amount of Nafion within the agglomerate. It is also found that platinum mass loading needs to be at a moderate level in order to optimize fuel cell performance, even if cost is to be disregarded. (C) 2008 Elsevier B.V. All rights reserved.