One hundred and fifty two patients with chronic heart failure caused primarily by left ventricular dysfunction were followed prospectively in an open study for a mean period of 21 months. The effects of several clinical variables on subsequent outcome were examined, including the effects of treatment, which was determined by the clinician caring for the patient and was not randomly allocated. In order of importance, frequent ventricular extrasystoles, non-treatment with amiodarone, low mean arterial pressure, and a diagnosis of coronary artery disease were associated with a poor prognosis, with each of these variables providing extra predictive information independently of the others. Initial serum potassium concentration and treadmill exercise time also carried further weak but independent prognostic information. Neither treatment with angiotensin converting enzyme inhibitors nor digoxin appeared to affect outcome. Left ventricular function (as reflected by M mode echocardiography) and the dose of diuretic also failed to predict outcome. There did, however, appear to be a reduction in the frequency of sudden death when angiotension converting enzyme inhibitors were given. Further studies are required to confirm the adverse prognostic significance of ventricular arrhythmias in patients with heart failure and the possible benefit associated with amiodarone treatment.