Theory predicts democracy should reduce corruption. Yet, scholars have found that while corruption is low at high levels of democracy, it is high at modest levels, as well as low when democracy is absent. A weakness of studies that aim to explain this inverted curvilinear relationship is that they do not disaggregate the complex concepts of democracy and corruption. By contrast, this paper disaggregates both. We demonstrate that the curvilinear relationship results from the collective impact of different components of democracy on different types of corruption. Using Varieties of Democracy data, we examine 173 countries from 1900 to 2015, and we find freedom of expression and freedom of association each exhibit an inverted curvilinear relationship with corruption—both overall corruption and four different types. The introduction of elections and the quality of elections each act in a linear fashion—positively and negatively with corruption, respectively—but jointly form a curvilinear relationship with both overall corruption and many of its types. Judicial and legislative constraints exhibit a negative linear relationship with executive corruption. We offer a framework that suggests how these components affect costs and benefits of engaging in different types of corruption and, therefore, the level of corruption overall.