Experience seems to suggest that government wide performance monitoring is more likely to prove effective when it is done in smaller, more homogeneous communities, in which agreement on appropriate outcome measures is easier to attain. Such schemes may also be more effective where the task of selecting outcomes is left to a body that is partly or wholly independent of government. At first, such an arrangement seems one of doubtful legitimacy, particularly if the outcome measures selected are expected to have a real influence on the shape of policy debate. However, an independent body is not constrained by some of the bureaucratic and political considerations that may make it difficult for governments to select a narrow group of outcome measures and report regularly on those measures. In the end, however, even a well-designed performance monitoring scheme may have only a limited impact on internal budgetary decision making or on popular opinion about government priorities.