This paper describes a mechanism for identifying the dynamics of non-ideal mixing processes. The object is to study two of the non-ideal behaviours of agitated pulp stock chests: recirculation and channelling. An initial continuous-time model, which contains physically relevant parameters, is transformed into its discrete-time counterpart. This transformation introduces some challenging identification problems, as the discrete-time parameters become a non-linear combination of the original continuous-time parameters. A system identification methodology that addresses these challenges is developed and demonstrated by means of computer simulation. The analysis of data collected from experiments on a laboratory scale model of an industrial chest shows the potential of the techniques developed in this paper. (C) 2004 Elsevier Ltd. All rights reserved.