The authors analysed survival of 160 patients (121 men and 31 women; average age 57.2 +/- 12.5 years; follow-up 29 +/- 20 months) treated for malignant ventricular arrhythmias (sustained ventricular tachycardia, ventricular fibrillation, syncope with inducible ventricular tachycardia). The therapeutic evaluation was frequently invasive (145 patients underwent at least programmed ventricular stimulation, 108 patients underwent full endocavitary electrophysiological studies) and non-pharmacological therapy was widely used (defibrillator n = 44; antiarrhythmic surgery n = 28; ablative procedures n = 19; transplantation n = 7). The following underlying pathologies were observed: ischaemic heart disease n = 120; non-ischaemic left heart disease n = 19; right heart cardiac disease n = 4; and apparently normal hearts n = 17). The average ejection fraction was 40.5 +/- 15.5% and 29 patients were in the NYHA functional classes III or IV. Fifty-five patients had life-threatening arrhythmias whilst receiving amiodarone. At 2 years, the actuarial sudden death rate was 5.9 +/- 2.1% and the actuarial total cardiac mortality rate was 13.1 +/- 2.9%. Univariate analysis showed age, the presence of underlying cardiac disease, the presence of dilated cardiomyopathy, the absence of an invasive approach, the need for basal pacing in electrical cardioversion, the absence of betablocker therapy, a decreased left ventricular ejection fraction and a high NYHA functional class, to be predictive of sudden death. In multivariate analysis, age, the NYHA class for total cardiac mortality and the NYHA class for sudden death, were the only independent predictive factors. The authors conclude that in the era of invasive methods of evaluation and widespread use of non-pharmacological therapeutic methods, the symptomatology of cardiac failure assessed by the NYHA classification remains the most powerful independent prognostic factor after an episode of malignant ventricular arrhythmia.