The policy preference function (PPF) approach has become popular with economists seeking to explain the origin of government policies. In this paper, a distinction between positive and normative work with the PPF concept is made. Positive work is shown to suffer from a variety of shortcomings including the misspecification of traditional PPFS and the failure to consider the importance of institutions, constraints and the interaction between different commodity policies. These weaknesses are reflected in the counter-intuitive results of a simple PPF model designed to reflect the interaction between the EC's wheat and barley policies. Furthermore, it is demonstrated that PPF weights change as a result of both political preferences and market parameters. Hence, changes in PPF weights cannot be attributed to changes in preferences alone. Tests of the axioms of revealed preference theory are used to demonstrate that even though PPF weights derived for the EC's wheat and barley markets have fluctuated considerably since the early 1970s, we are not able to conclude that there has been a shift in political preferences. The paper concludes with some comments about the use of PPFS in a normative framework. The underlying assumption that policy-makers optimise seems, not surprisingly, often to lead practitioners to the conclusion that observed policies are not so bad after all. Economists should also beware of the tendency to overlook possible differences between the PPF and the social welfare function.