A long-standing hypothesis states that family interaction mediates the course of schizophrenia. Over the last 30 years, Expressed Emotion (EE), a construct measured via assessments of family attitudes and behavior, and often found to predict schizophrenic relapse, has been central to research testing this hypothesis. This paper analyzes EE research, and concludes that conceptualization and measurement of EE are inadequate to clarify the schizophrenia-family interaction link. Behavioral assessment of the family interaction of schizophrenic patients has focused largely on the behavioral correlates of high EE. Relative to low-EE, high-EE families exhibit greater verbal negativity by both relatives and patients, coercive escalation, and higher autonomic arousal in patients. However, the association between these behavioral characteristics and the course of schizophrenia is largely untested. Research is needed which directly assesses the association between family interaction and the course of schizophrenia, which assesses outcome across a broad range of criteria, and which uses a wider range of available behavioral measures of family interaction.