Medicare spending varies more than twofold among regions, and the variations persist even after differences in health are corrected for. Higher levels of Medicare spending are due largely to increased use of "supply-sensitive" services--physician visits, specialist consultations, and hospitalizations, particularly for those with chronic illnesses or in their last six months of life. Also, higher spending does not result in more effective care, elevated rates of elective surgery, or better health outcomes. To improve the quality and efficiency of care, we propose a new approach to Medicare reform based on the principles of shared decision making and the promotion of centers of medical excellence. We suggest that our proposal be tested in a major demonstration project