A sizeable body of research has demonstrated that expressed emotion (EE) predicts clinical relapse in a number of distinct psychiatric disorders. These findings have provided the impetus for the development of interventions that attempt to reduce patients' relapse rates by modifying aspects of the family environment believed to be associated with high levels of EE. Despite the efficacy of these treatments, however, we know little about how EE develops in relatives of psychiatric patients or about the mechanisms through which high EE leads to relapse. Moreover, there is not a coherent theory that attempts to integrate findings concerning the impact of high EE on relapse in different disorders. The purpose of this article is to elucidate a diathesis-stress conceptualization of EE to explain both the development and manifestation of high EE in relatives of disordered patients and the impact of high EE on the course of patients' disorders. In this context, we use a diathesis-stress perspective to examine why EE predicts symptom relapse and poor clinical outcome in schizophrenia, depressive disorders, and borderline personality disorder. We conclude by discussing treatment implications of the diathesis-stress perspective and by outlining what we believe are fruitful directions for future research.