Mesoporous carbon is often used as a support for platinum catalysts in polymer electrolyte fuel-cell catalyst layers. Mesopores in the carbon support improve the performance of fuel cells by inhibiting the adsorption of ionomer onto the catalyst particles. However, the mesopores may impair mass transport. Hence, understanding molecular behaviors in the pores is essential to optimizing the mesopore structures. Specifically, it is crucial to understand the oxygen transport in the high-current region. In this study, the diffusion coefficients of oxygen molecules in carbon mesopores were calculated for various pore lengths, pore diameters, filling rates, and water contents in the ionomer via molecular dynamics simulations. The results show that oxygen diffusion slows by 2 orders of magnitude because of pore occlusion, and it slows down by an additional 1 or 2 orders of magnitude if ionomers are present in the pores. The occlusion can be theoretically predicted by considering the surface free energy. This theory provides some insight into mesoporous carbon designs; for instance, the theory suggests that narrow pores should be shortened to prevent occlusion. Slow diffusion in the presence of ionomers was attributed to the localization of oxygen at the dense ionomer-carbon interface. Thus, to improve oxygen transport properties, carbon surfaces and ionomer structures may be designed in such a manner as to prevent densification at the interface.