In the case of heterogeneous reactions, diffusion is the only mechanism, locally, of transport of species to and from a surface. Thus, accurate prediction of diffusive transport is a prerequisite for accurate prediction of the operation of devices in which heterogeneous reactions occur. Three different diffusion models are examined from the standpoint of both accuracy and efficiency. Two of these models, namely the dilute approximation (DA) model and the Schmidt number (SN) model, are approximate models, and are compared against a rigorous multi-component diffusion (MCD) model derived from the Stefan-Maxwell equation. Both hydrogen-air and methane-air combustion in a monolith channel are studied. Inlet equivalence ratio, Reynolds number (flow rate), and wall temperature are considered as parameters. The results show that both the DA model and the SN model are accurate within 2% irrespective of the equivalence ratio or fuel-the worst accuracy being for hydrogen combustion. The DA model and the SN model produce almost identical results. In comparison to the MCD model, the DA model is approximately twice as computationally efficient, while the SN model is 2-16 times more efficient. The accuracy and efficiency of the SN model, in conjunction with its simplicity, makes it an attractive choice for the treatment of diffusion in catalytic combustion calculations. (C) 2007 Elsevier Ltd. All rights reserved.