For a proper design of multicomponent adsorption systems, the availability of reliable adsorption models is paramount. In this work, an experimental and modelling analysis of PCE/TCE adsorption, aimed at investigating adsorption capacities in a binary system, is carried out. All the experimental tests are conducted at constant pH and temperature. The analyte concentration is varied over a wide range in order to cover as much as possible real cases that could lead to different adsorption behaviours and, consequently, to different modelling results. Experimental data show that the PCE adsorption capacity does not depend on TCE presence, and the equilibrium final conditions are not related to different analyte adsorption rates. The equilibrium adsorption data are analyzed by using different adsorption models, namely Langmuir multicomponent, Ideal Adsorbed Solution Theory (IAST) and Predictive Real Adsorbed Solution Theory (PRAST) models. The Langmuir model provides the worst data prediction since its basic hypotheses do not reflect the physical behaviour of the system. The IAST model does not provide a satisfactory prediction of binary data except for low liquid concentration levels, as it underestimates the PCE adsorption capacity and overestimates ICE. It follows that when the solid coverage increases, the PCE-TCE mixture behaviour is not ideal. Finally, the PRAST model, here developed for liquid systems for the first time, is considered in order to take into account a non-ideal system by including activity coefficients. The PRAST model provides a very good prediction of PCE adsorption experimental data but it is not useful to predict ICE adsorption data, its performance being even worse than IAST. The inability of the model to correctly predict both isotherms simultaneously lies in the particular type of non-ideality of the system itself, as shown by experimental data. (C) 2011 Elsevier B.V. All rights reserved.