In hydrocyclone, the separation involves a complex three-dimensional swirling flow. To improve the design and operation of hydrocyclones, it is important to investigate to flow profile and particle motion for enhanced separation. A three-phase computational fluid dynamics (CFD) model was developed to predict the flow patterns in the hydrocyclone, including prediction of the air core formation. The CFD method was used to simulate the flow fields inside a hydrocyclone to predict its separation efficiency. Three models, k-" model, Reynolds stress model (RSM), and Reynolds stress turbulence model with volume of fluid multiphase model for simulating air core, were compared for the predictions of velocity, axial, and tangential velocity distributions and separation efficiency. It is seen that the percentage of particles to underflow increases with particle size. Results indicate that the turbulence and discrete phase mode effects are clearly represents for predicting the experimental data with enhances separation efficiency.