Most models for the study of near-road air quality are based on either steady-state Gaussian equations or Lagrangian puff formulations to describe nonreactive pollutant concentrations at receptor sites. The accuracy of the dispersion calculation is often limited because of simplifications that are necessary to derive the analytical solutions. The general sourcereceptor structure of these models limits their capabilities to treat rapid chemical and physical processes that were recently observed in the near-road environment. A new three-dimensional (3-D) Eulerian near-road air quality model, the Texas A&M Near-Road Model (TAMNROM-3D), was developed. The enhanced pollutant turbulent diffusion caused by moving vehicles is handled in TAMNROM-3D by a revised vehicle-induced turbulence parameterization scheme. The performance of TAMNROM-3D on nonreactive tracer dispersion is evaluated with the SF6 data set collected by General Motors during a 1975 sulfate dispersion experiment. Predicted concentrations show good agreement with observations under different wind conditions. Model performance statistics are compared favorably with those reported by the ROADWAY-2 model. Sensitivity studies indicate that the model is robust to reasonable uncertainties in the model parameters and input data.