Natural ventilation is an energy saving system for the building to ensure occupant's physical comfort. It has also no contribution to the atmospheric pollution as well as the global warming. In the present study, a building with multiple windows is considered to investigate the wind-driven a ventilation system using computational fluid dynamics (CFD), whose acceptance and accuracy are growing very fast. The Renormalization group (RNG) k-epsilon turbulence model is chosen to simulate cross and single-sided ventilation with a specified accuracy after validating the methodology through the satisfactory comparison with an experimental result. The CFD model is then applied to investigate the physical mechanism of the air movement. The results are presented in the form of the mean velocity vectors, the magnitude of velocity, the components of velocity, the pressure distribution, the pressure coefficient and the effect of incoming wind velocity inside and outside the building. The necessity of three-dimensional (3D) approach to predict the indoor air movement correctly in studying natural ventilation system is also emphasized in this article. (c) 2011 Elsevier B.V. All rights reserved.