A two-dimensional, finite-difference model simulating a highway has been developed which is able to handle linear and nonlinear chemical reactions. Transport of the pollutants is accomplished by use of an upstream-flux-corrected algorithm developed at the Naval Research Laboratories. This algorithm insures positive concentrations and reduces numerical diffusion. The model develops background eddy diffusion coefficients which are then perturbed by the mechanical turbulence introduced by vehicular motion. The model calculations are compared to the SF6 experiments carried out by General Motors in the GM sulfate dispersion experiment. Model predictions are compared to observations at each of the twenty observation points for all the experiments. The observations and predictions are correlated for each half hour period, and scatter diagrams presented. A discussion of the results and additional statistical comparisons is also included.