We present a novel theory of mixture transport in nanopores, which represents wall effects via a species-specific friction coefficient determined by its low density diffusion coefficient. Onsager coefficients from the theory are in good agreement with those from molecular dynamics simulation, when the nonuniformity of the density distribution is included. It is found that the commonly used assumption of a uniform density in the momentum balance is in serious error, as is also the traditional use of a mixture center of mass based frame of reference. DOI: 10.1103/PhysRevLett.100.236103 PACS numbers: 68.43.Jk A fundamental understanding of the processes affecting the transport of fluid mixtures in nanoscale confinements is crucial to numerous emerging applications in nanotechnology, materials science, membrane science, and biology, as well as a host of other areas. For a long time, the modeling of mixture transport has relied on highly respected statistical mechanical theories, which relate the hydrodynamic stress tensor for any component to the rate of strain for the mixture motion as a whole [1,2]. Such theories involve expansion of the species velocities around the mixture center of mass velocity, but despite their rigor they have failed to provide satisfactory solutions to problems involving mixture transport over a wide range of densities. Indeed, there exists no definitive treatment even for a simple classical experiment known as the Stefan tube. The approaches have recently been criticized by Kerkhof and Geboers [3], who suggest expansions around the individual species center of mass velocities, as they can be very different from the mixture center of mass velocity. While also considered earlier by Snell, Aranow, and Spangler [4], such an expansion has not evoked much interest due to its complexity and the use of partial viscosities for which there is no simple prescription.For porous materials, perhaps the most commonly used approach is the dusty gas model [5], which also uses of the mixture center of mass as a frame of reference. Additionally, the approach neglects density gradients arising from the fluid-solid interaction and relies on the classical Poiseuille flow model for uniform fluids. Further, it neglects dispersive interactions, demonstrated by us [6,7] to be as much as an order of magnitude in error in estimating the low pressure diffusivity. It is the neglect of dispersive interactions which leads to the flux expression comprising additive viscous and diffusive terms, the latter including Knudsen (i.e., wall-mediated) diffusion. Thus, there exists no unambiguous way to introduce wall effects in the modeling. Finally, there is much confusion as to whether Onsager coefficients relate to the total flux or the diffusive component alone [8][9][10].Here we present a tractable theory that overcomes all of the above limitations and for the first time is able to handle mixture transport in nonuniform fluids from the nanopore to the mesopore range of confinement. We consider the one-dimensional axial ...