No one single model consistently out-performed the others in all performance objectives or measures and the state-of the art models (CALPUFF and AERMOD) did not exhibit superior performance in all performance objectives to the legacy models (ISC2 and RATCHET). Lagrangian puff models generally exhibited smaller variances, higher correlation, and higher percentage of predictions within a factor of two compared to the steady-state models at these distances. The conceptual framework of a Lagrangian puff model is better suited for long range transport where winds vary spatially across the model domain. Hence, Lagrangian puff models may be preferable for dose reconstruction where model domains can be large and where the assessment question is an unbiased estimate of concentration in time and space. However, model choice depends on site-specific considerations and the assessment questions to be addressed, and therefore no categorical statement can be made about the performance of one type of model over the other for a specific application.