This paper shows that the internal structure (relative sizes, spacings) of a fuel cell can be optimized so that performance is maximized at the global level. The optimization of flow geometry begins at the smallest (elemental) level, where the fuel cell is modelled as a unidirectional flow system. The polarization curve, power and efficiency are obtained as functions of temperature, pressure, geometry and operating parameters. Although the model is illustrated for an alkaline fuel cell, it may be applied to other fuel cell types by changing the reaction equations and accounting for the appropriate energy interactions. The optimization of the internal structure is subjected to fixed total volume. There are four degrees of freedom in the optimization, which account for the relative thicknesses of the two (anode and cathode) diffusion layers, two reaction layers and the space occupied by the electrolyte solution. The available volume is distributed optimally through the system so that the total power is maximized. Numerical results show that the optima are sharp, and must be identified accurately. Temperature and pressure gradients play important roles, especially as the fuel and oxidant flow paths increase. The optimized internal structure is reported in dimensionless form. Directions for future improvements in flow architecture (constructal design) are discussed. Copyright (C) 2004 John Wiley Sons, Ltd.