Rich democracies exhibit vast cross-national and historical variation in the socialization of health care. Yet, cross-national analyses remain relatively rare in the health policy literature, and health care remains relatively neglected in the welfare state literature. We analyze pooled time series models of the public share of total health spending for eighteen rich democracies from 1960 to 2010. Building on path dependency theory, we present a strategy for modeling the relationship between the initial 1960 public share and the current public share. We also examine two contrasting accounts for how the 1960 public share interacts with conventional welfare state predictors: the self-reinforcing hypothesis expecting positive feedbacks and the counteracting hypothesis expecting negative feedbacks. We demonstrate that most of the variation from 1960 to 2010 in the public share can be explained by a country's initial value in 1960. This 1960 value has a large significant effect in models of 1961–2010, and including the 1960 value alters the coefficients of conventional welfare state predictors. To investigate the mechanism whereby prior social policy influences public opinion about current social policy, we use the 2006 International Social Survey Programme (ISSP). This analysis confirms that the 1960 values predict individual preferences for government spending on health. Returning to the pooled time series, we demonstrate that the 1960 values interact significantly with several conventional welfare state predictors. Some interactions support the self-reinforcing hypothesis, while others support the counteracting hypothesis. Ultimately, this study illustrates how historical legacies of social policy exert substantial influence on the subsequent politics of social policy.