Managers and policymakers are seeking practical guidelines for assessing the outcomes of emerging pay-for-performance (P4P) programs. Evaluations of P4P programs published to date are mixed-some are confusing-and methodological problems with them are common. This article first identifies and summarizes obstacles to implementing effective P4P programs. Second, it describes results from social science research going back several decades to support evidence-based P4P best practices. Among the findings from this research, the zero-sum and "earn it back" P4P incentive systems have important drawbacks and may be counterproductive, neither reducing health system costs nor improving quality. The research suggests that punishing participants for low performance may further reduce individuals' performance, especially when involvement is required. We suggest that optimal P4P systems are those that reward all participants for performance improvements. Third, the article links P4P design to budgetary considerations. P4P program designs that provide incentives while improving quality and reducing costs are critical if budget neutrality is a priority for the organization and its resources are limited. In these tYpes of P4P designs, cost calculations are straightforward: The greater the participation, the higher the savings. The article concludes by recommending an evidence-based P4P approach for practitioners that can be implemented without large upfront investment. More research on this topic is also advised.