In an attempt to contain costs, 27 Medicaid programs have implemented patient-level payment limits for medications, but the effects of these restrictions on quality of care, costs, and health status remain largely unknown. We measured the effect of one state's limit of three paid prescriptions per month and its replacement a year later by a $1 copayment. Using data on 48 months of claims in the study state (New Hampshire) and a comparison state (New Jersey), we employed time-series analysis to evaluate patient-level changes in the number of prescriptions filled for 16 drugs that varied in their clinical importance and cost. Among 10,734 continuously enrolled patients, the limit of three paid prescriptions per month caused a sudden, sustained drop of 30 percent in the number of prescriptions filled (from 1.10 to 0.77 prescriptions per patient per month); no change was observed in the comparison state. The 860 recipients of multiple drugs, who were predominantly female and elderly or disabled, were most severely affected; the number of prescriptions per month dropped from 5.2 to 2.8 (46 percent). The decrease was greatest for "ineffective drugs" (58 percent), but large drops were also observed for "essential" medications, such as insulin (28 percent), thiazides (28 percent), and furosemide (30 percent). Reductions in Medicaid prescriptions were minimally offset by increases in the size of the prescription or in out-of-pocket payments. When a $1 copayment replaced the three-prescription cap, prescriptions for most medications increased to just below precap levels. Medicaid's savings on drug costs resulting from both policies were comparable ($0.4 to $0.8 million annually), but the copayment policy had less effect on patients receiving multiple drugs. Because the clinical consequences of such policies cannot be assessed from prescription data alone, further study is needed to determine the effects of cost-containment strategies on health status and the use of other services among poor populations.