A better understanding of tornado-induced wind loads is needed to improve the design of typical structures to resist these winds. An accurate understanding of the loads requires knowledge of near-ground tornado winds, but observations in this region are lacking. The first goal of this study was to verify how well a CFD model, when driven by far field radar observations and laboratory measurements, could capture the flow characteristics of both full scale and laboratory-simulated tornadoes. A second goal was to use the model to examine the sensitivity of the simulations to various parameters that might affect the laboratory simulator tornado. An understanding of near-ground winds in tornadoes will require coordinated efforts in both computational and physical simulation. The sensitivity of computational simulations of a tornado to geometric parameters and surface roughness within a domain based on the Iowa State University laboratory tornado simulator was investigated. In this study, CFD simulations of the flow field in a model domain that represents a laboratory tornado simulator were conducted using Doppler radar and laboratory velocity measurements as boundary conditions. The tornado was found to be sensitive to a variety of geometric parameters used in the numerical model. Increased surface roughness was found to reduce the tangential speed in the vortex near the ground and enlarge the core radius of the vortex. The core radius was a function of the swirl ratio while the peak tangential flow was a function of the magnitude of the total inflow velocity. The CFD simulations showed that it is possible to numerically simulate the surface winds of a tornado and control certain parameters of the laboratory simulator to influence the tornado characteristics of interest to engineers and match those of the field.