The relation between baseline clinical variables and subsequent mortality was examined in 295 patients with mild to moderate heart failure who participated in a multicenter trial comparing the effect on treadmill exercise tolerance of captopril, digoxin and placebo given in addition to a diuretic drug. At baseline study, all patients had a left ventricular ejection fraction less than or equal to 40%; 81% were in New York Heart Association functional class II. The etiology of heart failure was ischemic in 62% and nonischemic in 38%. During an average follow-up period of 16 months, 47 patients (16%) died and 24 deaths were classified as sudden. By univariate analysis, left ventricular ejection fraction, ventricular premature beat frequency, couplet frequency, ventricular tachycardia frequency, functional class, treadmill exercise time and nonischemic heart disease were statistically associated with mortality. With multiple logistic regression analysis, left ventricular ejection fraction was identified as the variable most closely associated with total mortality (p = 0.006). Twenty-seven percent of patients with an ejection fraction less than or equal to 20% died compared with 7% with an ejection fraction greater than or equal to 30%. Ventricular tachycardia frequency on Holter monitoring was independently associated with both total mortality (p = 0.008) and sudden death (p = 0.003). Patients with a ventricular tachycardia frequency of greater than 0.088 events/h had a mortality rate of 34% compared with 12% in those without ventricular tachycardia. In the multivariate model, functional class (p = 0.02) and etiology of nonischemic heart disease (p = 0.04) remained as independent predictors of mortality, whereas treadmill exercise duration did not.(ABSTRACT TRUNCATED AT 250 WORDS)