To compare the effect of thoracic epidural local anesthetic, epidural opioid, and intravenous opioid on pulmonary shunt fraction, arterial oxygenation, and hemodynamic changes during one-lung ventilation (OLV) in patients undergoing thoracic surgery. A prospective, randomized, double-blind study. A university hospital. Thirty-nine patients undergoing OLV for pulmonary resection. Patients were randomized into 1 of 3 groups: epidural bupivacaine (TEA-B group, n = 13), epidural sufentanil (TEA-S group, n = 13), or intravenous remifentanil (IV-R group, n = 13) during general anesthesia with propofol. A double-lumen tube was inserted, and mechanical ventilation with 100% oxygen was used in the lateral decubitus position. Hemodynamic variables and arterial and mixed venous blood gas analysis from the radial and pulmonary artery catheter were measured and shunt fraction was calculated during two-lung ventilation (TLV), 15, 30, and 60 minutes after the initiation of OLV, and 15 minutes after the reinstitution of TLV. Although mean arterial pressures 15 and 30 minutes after OLV in the IV-R group were significantly higher than the value in TEA-S group, cardiac output and pulmonary vascular resistance were maintained. Decreases in PaO(2), SaO(2), PvO(2), and SvO(2) and an increase in the shunt fraction after OLV were not different among groups and returned to baseline value after the resumption of TLV. Thoracic epidural bupivacaine, epidural sufentanil, and intravenous remifentanil-combined general intravenous anesthesia have comparable effects on shunt fraction and arterial oxygenation during OLV in patients undergoing thoracic surgery.