The configurational-bias Monte Carlo method, which is used for efficient generation of molecular models of n-alkane chains, is combined for the first time with the dual control-volume grand-canonical molecular-dynamics simulation, which has been developed for studying transport of molecules in pores under an external potential gradient, to investigate transport and separation of binary mixtures of n-alkanes, as well as mixtures of CO2 and n-alkanes, in carbon nanopores. The effect of various factors, such as the temperature of the system, the composition of the mixture, and the pore size, on the separation of the mixtures is investigated. We also report the preliminary results of an experimental study of transport and separation of some of the same mixtures in a carbon molecular-sieve membrane with comparable pore sizes. The results indicate that, for the mixtures considered in this paper, even in very small carbon nanopores the energetic effects still play a dominant role in the transport and separation properties of the mixtures, whereas in a real membrane they are dominated by the membrane's morphological characteristics. As a result, for the mixtures considered, a single pore may be a grossly inadequate model of a real membrane, and hence one must resort to three-dimensional molecular pore network models of the membrane. (C) 2004 American Institute of Physics.