This traces the historical decisions, concepts and key professional collaborations that laid the foundations for the formation of American psychiatric epidemiology during the 20th century, up to the discipline's institutional consolidation, circa 1980, when the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) was published. Thomas Kuhn's ‘disciplinary matrix’ is mobilized as a framework that allows the institutional and intellectual construction of a discipline to be analyzed as separate but intertwined components, without assuming that the two evolve in tandem. The identification of the strengths, as well as the frailties and internal divisions of the discipline as it developed, reveals a paradoxical situation: a time lag between psychiatric epidemiology's institutionalization and public recognition, on the one hand; and the weak coherence of its intellectual components, on the other hand. We briefly trace the origins of the split among the discipline's aetiological models of mental disorders and suggest that the lack of coherence among them has prevented psychiatric epidemiology from achieving the status of a normal scientific discipline, in the Kuhnian sense. Without more explicit attention to the intellectual rationale of the discipline, psychiatric epidemiology will continue to maintain a strong institutional dimension and weak intellectual matrix.