Cost-control incentives have become an important part of the health insurance landscape in the United States. These incentives are strongest in capitated managed care organizations, especially HMOs, because such organizations are paid a fixed amount regardless of the spending they generate. Using a sample of insurance claims from about 500 plans, I find that in HMOs, spending on anti-cholesterol drugs is 19% lower than in other insurance plans. This spending difference could reflect the strong cost-control incentives generated by capitation, but it could also reflect the selection of healthy, price-sensitive patients into HMOs, or it could simply reflect lower prices. To understand the difference, I estimate a model of physician prescribing and patient refill decisions. Patients in HMOs are nearly twice as sensitive to copays as are other patients, consistent with selection of healthy patients into HMOs. Even after adjusting for this selection, however, HMO physicians remain highly sensitive to drug procurement costs. This high sensitivity explains about a 20% of the spending difference; combined with lower prices, it explains about 25-55% of the difference. I find no evidence that these spending reductions result in fewer patient refills or the prescribing of less effective statins.