Background Left ventricular assist device (LVAD) failure and malfunction rates are critical gauges for establishing LVADs as a long-term therapy for end-stage heart failure patients. These device performance measures, however, have been inadequately characterized in the bridge-to-transplantation literature. Methods REMATCH is a randomized trial that compares optimal medical management with LVAD implantation for patients with end-stage heart failure. An independent committee adjudicated patient outcomes. The primary endpoint - survival - was analyzed by intention to treat using the log-rank statistic. Frequency of event occurrence was analyzed by Poisson regression. The time to first event was analyzed by the product limit method. Device performance was disaggregated into confirmed malfunctions and system failures. The latter were events in which patients could not be rescued with backup circulatory support measures. Results The 1-year survival rate was 52% (95% confidence limit [CL]; 40%-63%) for LVAD patients versus 28% (95% CL; 17%-39%) for medical patients and the 2-year survival rate was 29% (95% CL; 19%-40%) for LVAD patients versus 13% (95% CL; 5%-22%) for medical patients. System failure was 0.13 per patient per year and the confirmed LVAD malfunction rate was 0.90. Freedom from device replacement was 87% at 1 year and 37% at 2 years. Conclusions Despite the observed rates of device malfunction and replacement, LVAD implantation confers clinically significant improvement with regard to survival as compared with medical management. Device modifications and innovations for infection management exhibit great promise of improving device performance in the near future. © 2004 by The Society of Thoracic Surgeons.