During the past three decades the debates on cross-situational consistency and the analysis of person  situation interaction have developed largely in parallel, employing two distinct research paradigms. The fundamental theoretical questions addressed by both remain ultimately the same, but the dierences in the paradigms they have employed prevented ®ndings from one from bearing on the other. The goal of this article is to provide a unifying framework that bridges these two paradigms and helps resolve the `consistency paradox' by combining the advantages of both. Speci®cally, the framework clari®es the relationships between variance components and aspects of behavioral consistency such as cross-situational consistency, temporal stability, the stability of behavioral signatures, and the utility of forming various types of aggregate predictor. It also places competing accounts of low cross-situational consistency within a unifying framework, and shows that they are not contradictory but complementary facets of the general relationship between the organization of behavior and cross-situational consistency. These formal relationships and predictions are illustrated using the Carleton College data of .