Ultralow platinum loading and high catalytic performance at the membrane electrode assembly (MEA) level are essential for reducing the cost of proton exchange membrane fuel cells. The past decade has seen substantial progress in developing a variety of highly active platinum-based catalysts for the oxygen reduction reaction. However, these high activities are almost exclusively obtained from rotating disk electrode (RDE) measurements and have rarely translated into MEA performance. In this Review, we elucidate the intrinsic limitations that lead to a persistent failure to transfer catalysts’ high RDE activities into maximized MEA performance. We discuss catalyst-layer engineering strategies for controlling mass transport resistances at local catalyst sites, in the bulk of the catalyst layer and at the interfaces of the MEA to achieve high performance with ultralow platinum loading. We also examine promising intermediate testing methods for closing the gap between RDE and MEA experiments.