The performance of direct methanol fuel cells (DMFCs) can be significantly affected by the transport of methanol through the membrane, depolarising the cathode. In this paper, the literature on composite membranes that have been developed for reduction of methanol crossover in DMFCs is reviewed. While such membranes can be effective in reducing methanol permeability, this is usually combined with a reduction in proton conductivity. Measurements of methanol permeability and proton conductivity are relatively straightforward. and these parameters (or a membrane 'selectivity' based on the ratio between them) are often used to characterize DMFC membranes. However, we have carried out one-dimensional simulations of DMFC performance for a wide range of membrane properties, and the results indicate that DMFC performance is normally either limited by methanol permeability or proton conductivity. Thus use of a 'selectivity' is not appropriate for comparison of membrane materials, and results from the model can be used to compare different membranes. The results also show that Nafion (R) 117 has an optimum thickness, where DMFC performance is equally limited by both methanol permeability and proton conductivity. The model also indicates that new composite membranes based on Nafion (R) can only offer significant improvement in DMFC performance by enabling operation with increased methanol concentration in the fuel. A number of composite membrane materials that have been reported in the literature are shown to deliver significant reduction in DMFC performance due to reduced proton conductivity, although improved performance at high methanol concentration may be possible. (c) 2005 Elsevier B.V. All rights reserved.