Uncertainty in the transporting wind field is a major contributor to the uncertainty in an atmospheric dispersion prediction; the wind uncertainties include both turbulent velocity fluctuations and measurement/observational limitations. This makes comparison between dispersion model predictions and observed concentration values a difficult task, since we may not know how accurately a prediction can be made using a particular set of input data. Turbulence closure techniques provide a means of relating the wind field uncertainties to the dispersion prediction uncertainties, and therefore provide a basis for objectively characterizing a significant component of the overall model performance. The closure methodology and the requirements for use in model evaluation studies will be discussed.