A numerical model for gas-diffusion electrodes incorporating oxygen transport losses at the nanoparticle, agglomerate, and electrode scales as well as kinetic and ohmic polarizations was developed and used to simulate and optimize air electrodes for polymerelectrolyte fuel cells. The model correctly predicts trends in performance resulting from changes to the amount and distribution of platinum and relative humidity. The optimum platinum mass fraction in a Pt/C catalyst decreases as the total amount of platinum is reduced despite the concomitant increase in electrode thickness because the benefit to intra-agglomerate oxygen transport exceeds the detrimental impact on proton transport through the catalyst layer. (C) The Author(s) 2019. Published by ECS.