One hundred thirty-three patients with dilated heart failure, 80 with coronary artery disease, and 53 with idiopathic dilated cardiomyopathy were followed for a mean of 29 months. Patients with ischemic heart disease had a worse prognosis than those classified as having idiopathic cardiomyopathy. Features from history, physical examination, and diagnostic tests done when patients were referred to our clinic were checked for univariate association with survival and were used in Cox model survival analysis to define risk groups. Neither the overall group nor either subgroup showed a relationship between ejection fraction and survival. The best variables for predicting long-term mortality included underlying coronary artery disease, basal systolic blood pressure of less than 120 mm Hg, presence of congestion on chest radiogram, and age over 64. Other variables did not improve risk prediction in the overall group. Among patients with ischemic heart disease, blood pressure, congestion, maximal heart rate on treadmill test, and the presence of left bundle branch block on the initial electrocardiogram all contributed. Only systolic blood pressure and the symptom score were related to survival in idiopathic cardiomyopathy.