Legal scholars, economists, and political scientists are divided on whether voter initiatives and legislative referendums tend to produce outcomes that are more (or less) majoritarian, efficient, or solicitous of minority concerns than traditional legislation. Scholars also embrace opposing views on which lawmaking mechanism better promotes citizen engagement, registers preference intensities, encourages compromise, and prevents outcomes masking cycling voter preferences. Despite these disagreements, commentators generally assume that the voting mechanism itself renders plebiscites more democratic than legislative lawmaking. This assumption is mistaken. While it might seem unimaginable that a lawmaking process that directly engages voters possesses fundamentally anti-democratic features, this Article defends that very claim. To do so, this Article constructs a set of comparative analytical benchmarks based upon an assessment of the democratic features of representative legislatures and the anti-democratic features of appellate judiciaries, and employs those benchmarks to evaluate direct democracy. This analysis reveals two critical yet overlooked features of direct democracy: First, direct democracy incorporates many of the theoretical and practical difficulties associated with judicial review. Second, direct democracy risks outcomes that embed cycling preferences by eliminating voter choice over the policy making institution itself and over the range of policy matters combined for simultaneous negotiation. The core insight that emerges is not merely that common understandings about direct democracy are misguided; rather, it is that recognizing the anti-democratic features of direct democracy proves essential in determining the sorts of public policy questions that are, or are not, suitable to this form of policy making.