The Urban Dispersion International Evaluation Exercise (UDINEE) was designed to assess model capability in predicting the effects of a release from a radiological-dispersal device. Here, the dispersion of sulfur hexafluoride (SF6) releases performed during the Joint Urban 2003 field experiment, which took place in Oklahoma City, was simulated using urban-dispersion and urban-canopy models, with model performance evaluated by comparing modelled peak concentrations with measured concentrations over the course of nine intensive operating periods. Other metrics compared include the exposure duration, which is defined as the duration that the concentration exceeds 10% of the maximum concentration, and the SF6 cloud speed. Model performance varies widely across releases, with the urban-dispersion model having a higher percentage (43%) of maximum concentrations falling within a factor of two of the observed maximum concentration, compared with 30% for the urban-canopy model. With respect to the cloud speed, the results of the urban-canopy model are generally within a factor of two, but the urban-dispersion model typically overestimates cloud speeds by factors of two to six.