Abstract From March 1991 through July 1992, 1,001 patients having elective coronary artery bypass grafting were randomized to receive either continuous warm (≥35 °C) blood cardioplegia with systemic normothermia (≥35 °C) or intermittent cold (≤8 °C oxygenated crystalloid cardioplegia and moderate systemic hypothermia (≤28 °C). Preoperative variables including age, sex, prior coronary bypass grafting, hypertension, prior myocardial infarction, diabetes, angina class, and preoperative heart failure class were similar in both groups, as were the intraoperative variables of number of coronary grafts, mammary artery use, and cardiopulmonary bypass time. Aortic cross-clamp time was significantly longer in the warm group (46 ± 23 minutes versus 40 ± 21 minutes). Most postoperative variables including mortality (warm, 1.0%, and cold, 1.6%), Q wave infarction (warm, 1.4%, and cold, 0.8%), and need of an intraaortic balloon pump (warm, 1.4%, and cold, 2.0%) were similar between groups. Total neurologic events (warm, 4.5%, and cold, 1.4%; p < 0.005) and perioperative strokes (warm, 3.1%, and cold, 1.0%; p ≤ 0.02) were significantly higher in the warm group. Neurologic events included perioperative stroke (warm, 15 patients, and cold, 5 patients; p < 0.02), perioperative encephalopathy (warm, 2 patients, and cold, 1 patient), and delayed (≥3 in-hospital days) stroke (warm, 5 patients, and cold, 1 patient). All patients experiencing a stroke had a persistent neurologic deficit at the time of discharge. Encephalopathy resolved completely in all instances. Retrograde warm blood cardioplegia provides excellent myocardial protection that compares favorably with cold oxygenated crystalloid techniques. However, this large prospective, randomized trial strongly suggests that warm heart surgery is associated with a significantly increased rate of neurologic damage.