Twenty patients were studied during and after coronary artery bypass grafting (CABG). The patients were randomly assigned to two groups. In 10 patients, intraoperative general anesthesia (GA) was based on a combination of intravenous (IV) sufentanil and midazolam. In 10 other patients, thoracic epidural analgesia (TEA) with 0.375% bupivacaine plus sufentanil, 1:200,000 (5 micrograms/mL), and intraoperative GA with midazolam/N2O were used. During the prebypass period, lower heart rates (HR), less variability in mean arterial pressure (MAP), and systemic vascular resistance (SVR) was accompanied by less variability in plasma norepinephrine (NE) concentrations in the TEA group. Although no differences were observed between the groups in perfusion pressure and the amount of vasoactive drugs used during the bypass period, variability in SVR was less in the TEA group. This was accompanied by less increase in NE and less variability in epinephrine (E) plasma concentrations. However, cortisol (Co) release was higher during this period in the TEA group. Weaning from bypass was accompanied by higher MAP and SVR in the TEA group at a lower HR, whereas no differences were observed in the E, NE, and Co plasma concentrations between the groups. During the first and second postoperative days, better pain relief and lower E and Co plasma concentrations were found in the TEA group. It can be concluded that better hemodynamic stability during the prebypass and bypass periods was accompanied by less E and NE release during these periods. Co release in the TEA group was higher during the bypass period. Weaning from bypass was characterized by better hemodynamics in the TEA group.(ABSTRACT TRUNCATED AT 250 WORDS)