It is widely agreed that the increased polarization many countries experience is bad for democracy. However, existing assessments of how polarization affects democracy operate with simplified understandings of both polarization and democracy. Bringing empirical studies and democratic theory into dialogue, I argue that polarization cannot be understood as a single phenomenon that can be evaluated in one way. Moreover, its different kinds affect different parts of the democratic system in distinct ways. First, we must distinguish between the degree of polarization in a given context and the different kinds of polarization at play. Second, we must consider whether people have good reasons for their polarizing behavior or whether it is entirely irrational. If people have good reasons for their polarizing behavior, the problem lies elsewhere than in polarization itself. Third, we must distinguish between the content of polarized opinions and the process of opinion formation. Both can be assessed with democratic criteria, but they raise different questions. Finally, for democratic evaluation it matters where polarization occurs and thus, we must differentiate between different sites of polarization: civil society, election campaigns, and legislatures. I recommend a systemic approach to assessing the democratic implication of polarization, which analyzes both the effects of polarization at different sites and on democracy as a composite whole.