Biventricular pacing (BVP) improves hemodynamics and symptoms in patients with heart failure with bundle branch block. Patients with a left ventricular ejection fraction <0.35 and ventricular tachyarrhythmias are at risk of sudden cardiac death, and they benefit most from implantable cardioverter defibrillators (ICDs). No study has evaluated the efficacy of the BVP-ICD combination in patients with heart failure with no history of syncope or sustained ventricular tachycardia (VT) or ventricular fibrillation (VF). Our prospective, observational study was performed on 135 consecutive patients with heart failure (aged, 64 +/- 11 years; 76% male; New York Heart Association functional class, 3.1 +/- 0.8; ejection fraction 0.28 +/- 0.06; ischemic heart failure, 43%; QRS interval duration, 153 +/- 11 msec) treated at our cardiac pacing unit between January 1999 and April 2001. In the first year (control phase), BVP alone was implanted. After that, BVP with ICD backup was used (prophylactic phase). Follow-up data were obtained by outpatient visits with electrocardiographic and echocardiographic examinations done at 3-month intervals. For patients who died, we examined hospital records, death certificates, and autopsy reports. Follow-up time averaged 840 days. The first 47 patients received BVP alone. During follow-up study, 19% of these patients died suddenly, and 11% died of worsening heart failure. None of the patients who died suddenly had hemodynamic deterioration or BVP malfunction before the event. The BVP-ICD group comprised 88 patients (18% with VT/VF inducibility on electrophysiologic testing). During follow-up study, 32% of these patients (18% with positive electrophysiologic testing) had VT/VF episodes successfully treated by ICD; 5% received inappropriate discharges on atrial fibrillation; and 6% died of heart failure with 1 sudden cardiac death. Cox proportional hazards model in the BVP-ICD group compared with BVP alone revealed hazard ratios for all-cause mortality and sudden cardiac death of 0.76 (95% confidence interval [CI], 0.58 to 0.96; p = 0.01) and 0.08 (95% CI, 0.05 to 0.42; p <0.01), respectively, adjusting for baseline characteristics and follow-up duration. Mortality in patients with heart failure remains high after BVP implantation, mainly because of sudden cardiac death. Although there are limitations with an observational study, our experience suggests that ICD backup grants increased security in BVP patients without conventional class I ICD indications.