The realignment perspective has exerted an enormous amount of influence on thinking about American politics, but recently it has fallen into disfavor. As a theory of political change, this dissatisfaction is warranted. However, in rejecting the realignment perspective, scholars risk losing a valuable concept, the notion of a critical realignment. My thesis is that, properly defined, the concept of a critical realignment can be a powerful tool in the study of electoral behavior and an important component of a broader theory of political change. This thesis derives from an analysis of presidential elections between 1828 and 1984. This analysis provides dramatic evidence for the proposition that critical realignments are important electoral phenomena. The evidence is equally clear, however, that critical realignments are subnational phenomena that vary considerably in form, not the majestic national movements some believed them to be. The analyses reported here reveal broadly based electoral eruptions of 40 to 50 points that endure for decades.