Clinical guidelines, which increasingly build upon impartial analysis of evidence from well-designed studies, have become highly credible sources of information about what forms of care are effective. Consequently, they are attractive as foundations for performance incentives. Unfortunately, they are often complex, and frequently it is infeasible to gather the information required to assess compliance with guidelines at reasonable cost. I discuss the problems in implementing evidence-based guidelines and steps that could be taken to make them more useful as a basis for performance measurement.