The authors examine the design and limitations of incentives for health care providers to serve in rural areas in developing countries. Governments face two problems: it is costly to compensate well-trained urban physicians enough to relocate to rural areas, and it is difficult to ensure quality care when monitoring performance is costly or impossible. The goal of providing universal primary health care has been hard to meet, in part because of the difficulty of staffing rural medical posts with conscientious caregivers. The problem is providing physicians with incentives at a reasonable cost. Governments are often unable to purchase medical services of adequate quality even from civil servants. Using simple microeconomic models of contracts and competition, the authors examine questions about: a) The design of rural service requirements and options for newly trained physicians. b) The impact of local competition on the desirable level of training for new doctors. c) The incentive power that can be reasonably expected from explicit contracts. One problem a government faces is choosing how much training to give physicians it wants to send to rural areas. Training is costly, and a physician relocated to the countryside is outside the government's direct control. Should rural doctors face a ceiling on the prices they charge patients? Can it be enforced? The authors discuss factors to consider in determining how to pay rural medical workers but conclude that we might have to set realistic bounds on our expectations about delivering certain kinds of services. If we can identify reasons why the best that can be expected is not a particularly good, it might lead us to explore entirely different policy systems. Maybe it is too hard to run certain decentralized systems. Maybe we should focus on less ambitious but more readily achievable goals, such as providing basic infrastructure.