In a recent publication, J.G. Wijmans [J.G. Wijmans, Process performance = membrane properties + operating conditions, J. Membr. Sci. 220 (2003) 1-3] drew attention to the risk associated with a common practice in pervaporation: reporting experimental permeates fluxes without the corresponding operating conditions. The systematic use of the membrane permeability, in place of the raw fluxes, clearly improves the comparison as well as the prediction of process performances. Nevertheless, this strategy does not generally lead to a single value for the permeability coefficient; moreover, discrepancies are often reported for systems which exhibit variable permeabilities. This work analyses the limitations of the permeability concept based on a preliminary case study limited to a single penetrant (propane permeation through a silicone rubber membrane). In addition, it is shown that solution-diffusion model fundamentals enable experimental data reconciliation. This novel strategy is preferable when we wish to compare and compute the permeate flux on a rigorous basis. (c) 2005 Elsevier B.V. All rights reserved.