This article utilizes a theoretical framework of policy innovation, diffusion, and reinvention to investigate the evolving nature of Medicaid managed care programs over time. By estimating two separate models, one for primary care case management (PCCM) and a second for risk-based program enrollment, this study seeks to disentangle two different paths of learning (internal and external), investigate the potential effects of vertical diffusion of policy, and examine the impact of internal state characteristics on the extent of Medicaid managed care. With respect to diffusion and learning, the data reveal that earlier adopters implement more extensive programs. The data fail to reveal much internal learning, although there is evidence of some. External impacts are clear: managed care enrollments in neighboring states and changes in the federal waiver process affect states' decisions. Other policy choices are important: states with more generous Medicaid eligibility rules implement more extensive managed care programs. Complementing other studies of Medicaid, we find that politics and economics make a difference for the extent of managed care programs; unlike other Medicaid studies, we find no effect of race and ethnicity.