Increases in the activity of managed care organizations may have "spillover effects," influencing the entire health care delivery system's performance, so that care for both managed-care and non–managed-care patients is affected. Some proposals for Medicare reform have incorporated spillover effects as a way that increasing Medicare HMO enrollment could contribute to savings for Medicare.This paper investigates the relationship between HMO market share and expenditures for the care of beneficiaries enrolled in traditional fee-for-service Medicare. We find that increases in systemwide HMO market share (which includes both Medicare and non-Medicare enrollment) are associated with declines in both Part A and Part B fee-for-service expenditures. The fact that managed care can influence expenditures for this population, which should be well insulated from the direct effects of managed care, suggests that managed-care activity can have broad effects on the entire health care market. Increases in Medicare HMO market share alone are associated with increases in Part A expenditures and with small decreases in Part B expenditures. This suggests that any spillovers directly associated with Medicare HMO enrollment are small.For general health care policy discussions, these results suggest that assessment of new policies that would influence managed care should account not only for its effects on enrollees but also for its systemwide effects. For Medicare policy discussions, these findings imply that previous results that seemed to show large spillover effects associated with increases in Medicare HMO market share, but inadequately accounted for systemwide managed-care activity and relied on older data, overstated the magnitude of actual Medicare spillovers.