The trait-based perspective of leadership has a long but checkered history. Trait approaches dominated the initial decades of scientific leadership research. Later, they were disdained for their inability to offer clear distinctions between leaders and nonleaders and for their failure to account for situational variance in leadership behavior. Recently, driven by greater conceptual, methodological, and statistical sophistication, such approaches have again risen to prominence. However, their contributions are likely to remain limited unless leadership researchers who adopt this perspective address several fundamental issues. The author argues that combinations of traits and attributes, integrated in conceptually meaningful ways, are more likely to predict leadership than additive or independent contributions of several single traits. Furthermore, a defining core of these dominant leader trait patterns reflects a stable tendency to lead in different ways across disparate organizational domains. Finally, the author summarizes a multistage model that specifies some leader traits as having more distal influences on leadership processes and performance, whereas others have more proximal effects that are integrated with, and influenced by, situational parameters.