The Cardiac Arrhythmia Suppression Trial (CAST) database was analyzed with a Cox proportional hazards regression model to predict the mortality of patients with very poor left ventricular systolic function (ejection fraction < or = .20). Predictors of total death or cardiac arrest were (relative risk), QRS duration (1.10/10 msec increase), coronary artery bypass grafting (0.38), basal heart rate (1.26/10 min-1 increase), diastolic blood pressure (0.79/10 mm Hg increase), diabetes mellitus (1.59), EF (0.94/1 U increase), and ease of suppression (the ability to suppress ambient ventricular ectopy on the lowest dose of the first randomly chosen CAST drug) (0.64). Predictors of arrhythmic death or arrhythmic cardiac arrest included thrombolysis (0.44), coronary artery bypass grafting (0.38), diuretic use (1.71), heart rate (1.21/10 min-1 increase), calcium channel blocker use (1.69), and QRS duration (1.10/10 msec increase). Thus easily measurable clinical and laboratory variables help predict prognosis in this clinically important subgroup. The pathophysiologic basis for and the clinical implications of the ease of ventricular arrhythmia suppression correlating with prognosis requires further study.