A continuum modeling approach is used to compare the yield of catalyst particles called monodisperse, and containing only nanopores (i.e., micropores or narrow mesopores), to the yield of the same nanoporous particles to which larger diffusion channels have been introduced in various ways. Bidisperse and bimodal pore size distributions are considered; in the bimodal case, the large diffusion channels are allowed to vary in size throughout a particle, while in the bidisperse case they are all of the same size. Model-based optimization is performed with respect to the overall yield of a first-order reaction, optimizing the spatial distributions of both the volume fraction and the diameter of the diffusion channels. It is found that the yield of catalysts with an optimized bidisperse pore size distribution may be at least an order of magnitude increased with respect to the original monodisperse distribution, thanks to the optimized introduction of large pore channels. Remarkably, however, local variations in pore diameter and porosity in bimodal pore networks do not lead to significant further improvements. This calls for the synthesis of bidisperse catalysts with a carefully optimized size of the large pore channels and thickness of their nanoporous catalytic walls. (c) 2007 Elsevier Ltd. All rights reserved.