Using redox enzymes as biocatalysts in fuel cells is an attractive strategy for sustainable energy production. Once hydrogenase for H2 oxidation and bilirubin oxidase (BOD) for O2 reduction have been wired on electrodes, the enzymatic fuel cell (EFC) thus built is expected to provide sufficient energy to power small electronic devices, while overcoming the issues associated with scarcity, price and inhibition of platinum based catalysts. Despite recent improvements, these biodevices suffer from moderate power output and low stability. In this work, we demonstrate how substrate diffusion and enzyme distribution in the bioelectrodes control EFC performance. A new EFC was built by immobilizing two thermostable enzymes in hierarchical carbon felt modified by carbon nanotubes. This device displayed very high power and stability, producing 15.8 mW h of energy after 17 h of continuous operation. Despite the large available electrode porosity, mass transfer was shown to limit the performance. To determine the optimal geometry of the EFC, a numerical model was established, based on a finite element method (FEM). This model allowed an optimal electrode thickness of less than 100 μm to be determined, with a porosity of 60%. Thanks to very efficient enzyme wiring and high enzyme loading, non-catalytic signals for both enzymes were detected and quantified, enabling the electroactive enzyme distribution in the porous electrode to be fully determined for the first time. High total turnover numbers, approaching 107 for BOD and 108 for hydrogenase, were found, as was an impressive massic activity of 1 A mg−1 with respect to the mass of the electroactive enzyme molecules. This strategy, relying on stable enzymes and mesoporous materials, and the model set up may constitute the basis for a larger panel of bioelectrodes and EFCs.