In this article, the author discusses the use in policy analysis of models that incorporate uncertainty. He believes that all models should consider incorporating uncertainty, but that at the same time it is important to understand that sampling variability is not usually the dominant driver of uncertainty in policy analyses. He also argues that there is no reason to avoid the use of simulation, as it is a powerful tool for analyzing the uncertain links that connect things that have been studied empirically with the quantities of interest in a policy analysis. The real challenge, he says, is to communicate effectively about uncertainty to the relevant audience. (Contains 2 footnotes.)