MIRA and Rover Group Ltd have undertaken a systematic study of the ability of CFD methods to predict the aerodynamic characteristics of simplified car-like shapes. This paper reports the latest stage of this investigation, which examines the use of a commercial Reynolds-averaged Navier-Stokes code (STAR-CD) to predict the aerodynamic characteristics of a series of simplified car shapes.Comparable experimental data were obtained by testing in the MIRA Full Scale Wind Tunnel (FSWT).This paper shows that CFD techniques are improving in their ability to predict flow separation from curved surfaces accurately. Further, encouraging results for vehicle drag (coefficients to within 2% of experiment) and the effect of limited geometric modifications on drag (within 7% of full-scale experiment) were obtained. However these latter results should be viewed with some caution as the results for lift were considerably poorer. Nevertheless, the utility of two-layer turbulence modeling is demonstrated along with the vital importance of using spatial discretisation schemes of at least second-order accuracy.