Models for predicting short-term NO2 concentrations are discussed, and several (RPM-II, TCM, OLM, and CNOM) are selected for evaluation. The MISTT data, collected in 1976, were to be used to evaluate the models, but careful scrutiny of the data base revealed certain deficiencies relative to the data needs of the models. These deficiencies preclude a strict evaluation of the performance of the models, but simple fitting techniques were used to compensate for input data deficiencies. The models are shown to perform reasonably well using simple statistical measures of performance. The performance of the models is also evaluated using a 'restricted' data base (i.e., one that could be derived from NSW, local, or state agency data sources only), and model performance is shown to be poorer with the 'unrestricted' data base.