Pollutant measurements in traffic tunnels have been used to estimate motor-vehicle emissions for several decades. The objective in this type of study is to use the traffic tunnel as a tool for characterizing motor vehicles rather than seeking a tunnel design with acceptably low pollutant concentrations. In the past, very simple aerodynamic models have been used to relate measured concentrations to vehicle emissions. Typically, it is assumed that velocities and concentrations are uniform across the tunnel cross section. In the present work, a vehicle emitting a known amount of sulfur hexafluoride (SF6) was driven repeatedly through a 730-m-long traffic tunnel in Vancouver, Canada. Comparing the measured SF6 concentrations to the known emission rates, it is possible to directly assess the accuracy of the simple tunnel aerodynamic models typically used to interpret tunnel data. Correction factors derived from this procedure were then applied to measurements of carbon monoxide and other pollutants to obtain gram-per-kilometer emission factors for vehicles. Although the specific correction factors measured here are valid only for the tunnel tested, the magnitude of the factors (up to two or more) suggests that the phenomena observed here should be considered when interpreting data from other tunnels.