Since the mid-1980s, health maintenance organizations (HMOs) have grown rapidly in the United States. But despite initial successes in constraining health care costs, they have come under increasing criticism for their restrictive practices. This suggests that, to remain viable, HMOs must change their behavior. Yet few studies offer empirical evidence on the matter. The present study investigates one cost-containment mechanism often associated with HMOs: the assignment of primary care physicians as gatekeepers (who, among other things, monitor patients' use of specialist physicians). In particular, we estimate the effect of physician-HMO involvement on the percentage of HMO patients for whom physicians serve as gatekeepers. We examine this relationship over two time periods: 2000–2001 and 2004–2005. Because physicians can choose whether and to what extent they participate in HMOs, we employ instrumental variables (IV) estimation to correct for the endogeneity of the HMO measure. Although the single-equation estimates suggest that HMO assignment of physician gatekeepers diminished modestly over time, the endogeneity-corrected estimates show no change between the two time periods. Thus, one major tool used by HMOs to constrain health care costs—the physician gatekeeper—has not declined even in a period of backlash against managed care.