The predictive validity of expressed emotion (EE) and two conceptually related but more easily measured alternatives--marital distress, and patients' perceptions of criticism from spouses--were examined in a sample of hospitalized unipolar depressives. All three psychosocial variables were significantly associated with 9-month relapse rates. Expressed emotion and marital distress predicted the same proportion of variance in patients' outcomes. The single best predictor of relapse, however, was a patient's response to the question "How critical is your spouse of you?" Patients who relapsed rated their spouses as significantly more critical than did patients who remained well. Alone, the perceived criticism variable accounted for more of the variance in relapse rates than that explained by EE and marital distress combined. The results suggest that asking depressed patients how critical they believe their relatives are may facilitate the identification of individuals at high risk for relapse subsequent to hospital discharge.