This dissertation describes and explains the development of parliamentarism in Western Europe. Defining parliamentarism as an institutional solution in which the government is politically responsible to parliament only, I use country historiography to map out a tug-of-war between parliament and the head of state over the ability to make governments resign or maintain them in office in 11 West European countries since the establishment of national parliaments. To describe the development of parliamentarism, I use a Bayesian learning model that estimates how expectations of who might make the government in office resign are updated, based on information on previous government-termination attempts at each point in time. I argue that parliamentarism is institutionalized when past experience suggests that there is good reason to believe that no actor other than parliament can make the government resign. In addition, I theorize that the emergence of party systems, and the development of party-system fragmentation and polarization, affect the ability and willingness of parliamentary party leaders and heads of state to enforce parliamentarism or its counter-factual, power sharing, in practice. Having described the development of parliamentarism in the countries included in this dissertation, I show that parliamentarism never developed before the emergence of party systems, and I find support for my theory in time-series cross-sectional regression. Party-system fragmentation affects parliamentarism negatively, and some party-system polarization affects parliamentarism positively while much party-system polarization affects parliamentarism negatively. I complement the regression analyses by analyzing the behavior of the relevant actors in four case studies: Denmark from the 1850s to the 1920s, Belgium from the 1830s to the 1950s, France from the 1940s to the 1960s, and Finland from the 1920s to the 1990s. These case studies substantiate my argument about the effect of party systems further. Thus, the dissertation illustrates how a Bayesian learning model can be used to estimate institutional change and contributes with substantive knowledge about the development of a very important political institution in Western Europe. The results have implications for knowledge about West European democratization, the role of parties to enforce political institutions, and understandings of how political conflict can have both positive and negative effects on politics.