In the context of a standard model of optimal monetary policy, I argue that expectations should be treated as adaptive rather than rational. This argument is justified by considering the rational expectations equilibrium of this model as the limit point of a sequence in which agents progressively modify their forecasts of inflation to make them efficient. I show that this learning process is unlikely to occur, in real time, because of the large amount of data that would be required. When expectations are adaptive, there is no longer a time-inconsistency problem, and since inflation policy influences expectations of future inflation, the central bank s concern for its reputation induces it to deliver optimal (time-consistent) policy. In a final section, the implications of these results for central bank independence are discussed.