In recent years, many computational fluid dynamics (CFD) studies have appeared attempting to predict cyclone pressure drop and collection efficiency. While these studies have been able to predict pressure drop well, they have been only moderately successful in predicting collection efficiency. Part of the reason for this failure has been attributed to the relatively simple wall boundary conditions implemented in the commercially available CFD software, which are not capable of accurately describing the complex particle-wall interaction present in a cyclone. According, researches have proposed a number of different boundary conditions in order to improve the model performance. This work implemented the critical velocity boundary condition through a user defined function (UDF) in the Fluent software and compared its predictions both with experimental data and with the predictions obtained when using Fluent’s built-in boundary conditions. Experimental data was obtained from eight laboratory scale cyclones with varying geometric ratios. The CFD simulations were made using the software Fluent 6.3.26.