The cost of separating binary mixtures by distillation is compared with that of separation by membranes. The comparisons are based on design studies for mixtures of varying molecular weight and relative volatility. These were chosen to represent a wide range of conditions from cryogenic to vacuum distillation. The cost estimates for the membrane separators are based on recently published correlations of costs predicted for commercial scale plant. The comparisons are biased in favour of membranes and the use of 'targeting' enables the study to include hypothetical membranes as well as existing ones. The results explain why polymeric membranes, now used for the separation of hydrogen or water vapour from gas mixtures and for some air separations, are unlikely ever to be used to separate those mixtures which are now separated by conventional distillation. The potential advantages of molecular sieve membranes of high permeability and selectivity are such that, were they ever made available on a commercial scale, they might replace distillation for some separations.