In recent years, many health maintenance organizations (HMOs) have exited Medicare+Choice (M+C), the program that provides a managed-care option to Medicare. This paper answers the following questions: How does the equilibrium number of HMOs participating in county M+C markets vary with the capitation payment they are offered? How large a payment is required at the margin to ensure that various percentages of county markets have a M+C HMO, or to ensure that various percentages of Medicare beneficiaries have the choice of a M+C plan in their county of residence? The strategy for identifying the effect of government payment on HMO participation relies on a natural experiment; in 1997, Congress divorced M+C payments to HMOs from changes in underlying costs. The results in this paper suggest that the Centers for Medicare & Medicaid Services (CMS) has consistently underestimated the payment necessary to support HMOs in rural, sparsely populated areas. We also find that it would require a large incremental payment to support HMOs in M+C for the final 10% of counties or final 10% of Medicare beneficiaries.