The fish-hook effect is one of the most pronounced mechanisms responsible for air classification. In this work, fish-hook phenomenon vis-a-vis particle separation in a circulating air classifier is analyzed. The work is focused on systematic performance evaluation of circulating air classifiers. An attempt has been made to describe the effect of process and design parameters on the shape of the fish-hook. In this study, it is found that the increase in bypass leads to a remarkable fish-hook phenomenon. Based on the experimental data, a process model is developed to predict the bypass fraction within the classifier. The CFD simulation results are interpreted to understand the fish-hook behavior. It is conclusively deduced by analyzing the results of CFD simulations, that the fish-hook effect is due to wall rebounding. The proposed mechanism and the process model can be used to predict the underlying phenomena and also for selection and design of an air classifier. (C) 2011 Elsevier B.V. All rights reserved.