Health care payment models that reduce costs by changing physician behavior are the current trend in US health care reform. The uneven implementation of these reforms, however, leaves scope for strategic behavior on behalf of doctors. Is paying doctors to reduce costs an effective cost control strategy when only implemented in some hospitals and for certain patients? I look at a cost reduction policy where hospitals gave doctors bonuses to reduce costs for admitted Medicare patients. My model of physician decision-making generates three predictions: (1) for admitted patients, healthier patients will be sent to the hospitals offering cost reduction bonuses; (2) admission will go up in participating hospitals, as only admitted patients are eligible for the bonuses; (3) the bonus program will cause doctors to perform fewer services. I look at doctors who practice in both participating and non-participating hospitals and examine whether they change their behavior in the participating hospitals after the program is introduced. I find strong empirical support for the first two predictions but not the third; doctors sorted patients and increased admission, but did not lower costs. These results have important implications for health care reform. Previous research has shown that doctors respond as expected to financial incentives when looking at the narrow choice between competing procedures. This paper takes looks at the effect of financial incentives on each link of the decision-making chain, and finds that all are important. Furthermore, I show that losses from strategic behavior may outweigh gains from reduced costs down the line.