Pervaporation is proved to be a commercially viable membrane separation process by this time. However, to become a widely applied process in the industry it is of crucial importance to develop membrane properties, process and module design as well as proper modelling in professional software environment. In this work a pervaporation model improvement of the basic solution-diffusion model (Rautenbach et al., 1990) is recommended and tested on experimental data. The reason behind this improvement is that the transport coefficient cannot be considered as constant assumed in the basic model in a wide concentration range. The change of the transport coefficient is considered as an exponential function of the composition of permeating compound. This exponential dependency is assumed by the authors after investigating the shape of the flux curves measured in a wider feed concentration range. The accuracy of this improved model is experimentally tested with the pervaporation of isobutanol-water, n-butanol-water, and ethanol-water mixtures on commercial hydrophilic composite membranes. This model improvement gives accurate and reliable data for a wide range of feed concentration proving that the assumption of practically constant transport coefficient cannot be applied. Therefore the use of this improved model in professional flowsheeting software packages is more reliable and reasonable than the application of the basic model for the design and operation of pervaporation, or a more effective hybrid separation system including pervaporation. (C) 2012 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.