We have evaluated the performance of several urban Atmospheric Transport and Dispersion (AT&D) models by comparing model predictions to tracer gas concentrations measured during the Joint Urban 2003 field experiment in Oklahoma City, USA. These models include the Urban Canopy, Urban Dispersion Model (UDM), and Micro-SWIFT/SPRAY (MSS) modes within the HPAC modelling suite, QUIC-URB/QUIC-PLUME models, and the MESO/RUSTIC models. We discuss some of the results of these comparisons, including relative model performance according to bias and scatter metrics, differences in model behaviour for predictions of daytime vs. nighttime releases, and operational considerations such as runtime differences.