We performed an open, prospective, randomized, controlled study of the incidence of major organ complications in 420 patients undergoing routine coronary artery bypass graft surgery with or without thoracic epidural anesthesia and analgesia (TEA). All patients received a standardized general anesthetic. Group TEA received TEA for 96 h. Group GA (general anesthesia) received narcotic analgesia for 72 h. Both groups received supplementary oral analgesia. Twelve patients were excluded-eight in Group TEA and four in Group GA-because of incomplete data collection. New supraventricular arrhythmias occurred in 21 of 206 patients (10.2%) in Group TEA compared with 45 of 202 patients (22.3%) in Group GA (P = 0.0012). Pulmonary function (maximal inspiratory lung volume) was better in Group TEA in a subset of 93 patients (P < 0.0001). Extubation was achieved earlier (P < 0.0001) and with significantly fewer lower respiratory tract infections in Group TEA (TEA = 31 of 206, GA = 59 of 202; P = 0.0007). There were significantly fewer patients with acute confusion (GA = 11 of 202, TEA = 3 of 206; P = 0.031) and acute renal failure (GA = 14 of 202, TEA = 4 of 206; P = 0.016) in the TEA group. The incidence of stroke was insignificantly less in the TEA group (GA = 6 of 202, TEA = 2 of 206; P = 0.17). There were no neurologic complications associated with the use of TEA. We conclude that continuous TEA significantly improves the quality of recovery after coronary artery bypass graft surgery compared with conventional narcotic analgesia.