The effects of thoracic epidural anesthesia (TEA) on total body oxygen supply-demand ratio are complex due to potential influences on both O2 delivery (QO2) and consumption (VO2). One hundred and five patients undergoing abdominal aortic surgery were randomly assigned to one of three groups to compare the cardiovascular and metabolic responses associated with (1) thoracic epidural anesthesia plus light general anesthesia (group TEA); (2) general anesthesia with halothane (group H); and (3) neuroleptanalgesia (group NLA). Values of cardiac index (CI) and QO2 were less intraoperatively in the TEA group than in the H or NLA groups, while VO2 values were similar. VO2 during recovery was greater in both the TEA and NLA groups than in the H group. Consequently the oxygen supply-demand ratio (QO2/VO2) was less in the TEA group throughout the perioperative period and about 30% below baseline values during early recovery. At comparable VO2, CI and mixed venous O2 saturation were always less in the TEA group than in the NLA group. Heart rate was slowest intraoperatively during TEA, and stroke work was less with TEA than with NLA. As cardiac filling pressure and systemic vascular resistance did not differ among the three groups, reduced adaptation of CI to tissue O2 needs during TEA was attributed to negative inotropic and chronotropic effects of the sympathetic blockade. We conclude that in patients undergoing abdominal aortic surgery, TEA has no apparent advantage over general anesthesia.