This paper develops a decision-theoretic approach to policy analysis. We argue that policy evaluation should be conducted on the basis of two factors: the policymaker's preferences, and the conditional distribution of the outcomes of interest given a policy and available information. From this perspective, the common practice of conditioning on a particular model is often inappropriate, since model uncertainty is an important element of policy evaluation. We advocate the use of model averaging to account for model uncertainty and show how it may be applied to policy evaluation exercises. We illustrate our approach with applications to monetary policy and to growth policy. (This abstract was borrowed from another version of this item.)