Different forms of physician payment result in different levels of financial risk for health care payers and providers, and can affect clinical decision making and the cost of care. Shared savings programs reward providers for holding spending below specific targets, thus introducing a level of financial accountability not present in strictly volume-based payment models, such as fee-for-service. Here we examine the design and application of shared savings formulas across a range of actual programs. We also present a more detailed description of one particular shared savings program-the Massachusetts Patient-Centered Medical Home Initiative-focusing on key trade-offs between payers and providers that eventually led to agreement on specific aspects of the program. We conclude with principles for the design of future shared savings arrangements and consideration of issues that will confront decision makers as these efforts mature and expand.