Our study investigated the effects of moderate doses of fentanyl and sufentanil versus high-dose sufentanil on cerebral hemodynamics by using transcranial Doppler ultrasonography (TCD). Thirty American Society of Anesthesiologists (ASA) II and III patients scheduled for elective coronary artery bypass graft (CABG) were studied after Institutional Review Board (IRB) approval and informed consent. The evening before surgery, all patients received oral flurazepam (1 mg/kg), Atropine (0.4 mg/70 kg s.c.) and a combination of droperidol (70 micrograms/kg s.c.) plus fentanyl (1.5 micrograms/kg s.c.) were given as preanesthetic medication 1 h before induction of anesthesia. Anesthesia was induced with either 25 micrograms/kg fentanyl i.v. (group 1, n = 10), 3 micrograms/kg sufentanil i.v. (group 2, n = 10) or 6 micrograms/kg sufentanil i.v. (group 3, n = 10). All patients received 100 micrograms/kg pancuronium i.v. With the induction of respiratory depression, assisted ventilation was performed followed by controlled ventilation to maintain normoxia and normocapnia (FiO2, 1.0). Cerebral blood flow velocity (CBFV, cm/s) was measured continuously in the middle cerebral artery by using a bidirectional 2-MHz TCD system. Monitoring included heart rate (HR, beats/min), direct mean arterial blood pressure (MAP, mm Hg), and PaCO2. Physiologic variables including arterial blood gases were measured at baseline, 5 min, and 10 min after infusion of fentanyl or sufentanil. In all patients, HR, MAP, end-tidal carbon dioxide tension (PetCO2), and PaCO2 were constant over time and did not differ between groups. CBFV did not change with moderate doses of fentanyl (group 1) or sufentanil (group 2). In contrast, infusion of high-dose sufentanil (group 3) was associated with 27 to 30% decreases in CBFV (p < 0.05). Our results suggest that sufentanil decreases CBFV in a dose-related fashion with a threshold effect. Increases in CBFV and CBF seen in previous studies may be related to an increasing PaCO2 when maintenance of normocarbia is based on only real-time capnography with a constant PetCo2 rather than additional arterial blood gas monitoring.