The functionality doctrine in trademark law bars protection for some, but not all, sourceidentifying product features-so-called trade dress-that contribute to a product's functional performance. Despite the doctrine's lengthy history, its critical role in promoting intellectual property policies, and the considerable attention devoted to it in recent decades, courts and commentators still disagree about what functionality means, the reasons why functional marks should not be protected, and how far the functionality bar should extend. This confusion is due largely to a lack of clarity and rigor at the normative level. This article seeks to remedy the deficiency. It traces the history of the functionality doctrine, critically analyzes its policy foundations, and outlines an analytical approach for designing optimal functionality rules.