The National Center for Environmental Prediction’s (NCEP) Eta model is used for operational forecasts over the U.S. with a grid spacing of 32 or 12 kilometers. Advances in computer hardware will soon permit operational forecasts for limited-area local domains with a horizontal resolution much less than 10 km. Determining mesoscale model performance at smaller scales becomes problematic because there are usually insufficient routine observations made that resolve local boundary layer circulations, especially in mountainous regions. One source of forecast error in mountainous areas can be attributed to the vertical coordinate system employed by mesoscale models. Some models use a ‘sigma’, or terrain-following, coordinate system in which the pressure gradient force consists of two terms where both terms are large over steep slopes. It has been known for decades that an inconsistent discretization of these terms can lead to errors in the horizontal momentum equations. The ‘step-mountain’ vertical coordinate, however, eliminates these discretization errors. In this study, we compare Meso Eta model simulations that employ either the ‘sigma’ and ‘step-mountain’ coordinate system to examine the effect of vertical coordinate on the predicted boundary layer circulations in the Salt Lake City Valley. Running the same code with different coordinate systems eliminates the possibility that the differences in simulation results are due to model physics and parameterizations. A horizontal grid spacing of about 0.8 km is used for a domain that encompasses the Salt Lake Valley and the surrounding mountains and valleys. The Meso Eta simulations are evaluated using VTMX field campaign observations made during October of 2000. Simulated thermally-driven flows influenced by terrain from both coordinate systems are compared with measurements from surface meteorological stations, radiosondes, radar wind profilers, and other instrumentation in the valley. A statistical and qualitative evaluation of the performance of each coordinate system is made.