BACKGROUND: Progression of heart failure can lead to cardiac transplantation, but when patients are ineligible, long-term mechanical circulatory support may improve survival. The REMATCH trial showed that left ventricular assist devices (LVADs) prolonged survival in patients with end-stage disease, but with a significant number of adverse events. We report on the neurological outcomes in the REMATCH trial. METHODS AND RESULTS: We examined new neurological events in the 129 patients randomized to either LVAD placement (n=68) or medical management (n=61), classified as stroke, transient ischemic attack, toxic-metabolic encephalopathy, and other. There were 46 neurological events: 42 in 30 LVAD patients and 4 in 4 patients in the medical arm (chi2, 30/68 versus 4/61, P<0.001). Sixteen percent of the LVAD patients had a stroke, with a rate of 0.19 per year (95% CI, 0.10 to 0.33), many occurring in the postoperative period. The stroke rate in the medical arm was 0.052. A Kaplan-Meier survival analysis showed a 44% reduction in the risk of stroke or death in the LVAD group versus the optimal medical group (P=0.002). The mean interval from implantation to stroke was 221.8 days (+/-70.4 days). History of stroke, age, and sepsis were not stroke risk factors in the LVAD group. CONCLUSIONS: Fewer than half of the patients in the LVAD group had a neurological event, and there were few neurological deaths. Survival analysis combining stroke or death demonstrated a significant benefit for long-term circulatory support with an LVAD over medical therapy. Future trials will need to address prospectively all neurological outcomes, including neurocognitive function, and the role of long-term neuroprotection.