A method based on transfer function matching is used to obtain the backmixing coefficient of a solvent extraction column. The method demonstrates the exploitation of the symbolic manipulation capability of the current generation of scientific computing software. It starts with the dynamic model of Dongaonkar et al. and makes use of a set of tracer pulse data of Dongaonkar for a Kuhni column. Compared to the existing methods of evaluating backmixing coefficient, this new method is more direct and simple to implement on computers. It has the potential of being applicable to more elaborate dynamic models of plate columns. (c) 2005 Elsevier Ltd. All rights reserved.