We studied 80 healthy ASA I patients (aged 20-52 yr) to determine if fentanyl affects sevoflurane requirements for achieving 50% probability of no movement in response to laryngoscopy and tracheal intubation (MAC-TI). Patients were allocated randomly to one of four fentanyl dose groups (0, 1, 2 and 4 mu g kg(-1)). Patients in each group received sevoflurane at a pre-selected end-tidal concentration according to an 'up-down' technique. After steady state sevoflurane concentration had been maintained for at least 10 min, fentanyl was administered i.v. Tracheal intubation was performed 4 min after administration of fentanyl, and patients were assessed as moving or not moving. Heart rate (HR) and mean arterial pressure (MAP) were recorded before induction of anaesthesia, just before administration of fentanyl, just before laryngoscopy for intubation, and after intubation. The MAC-TI of sevoflurane was 3.55% (95% confidence intervals 3.32-3.78%), and this was reduced markedly to 2.07%, 1.45% and 1.37% by addition of fentanyl 1, 2 and 4 mu g kg(-1), with no significant difference in the reduction between 2 and 4 mu g kg(-1), showing a ceiling effect. Fentanyl attenuated haemodynamic responses (HR and MAP) to tracheal intubation in a dose-dependent manner, even with decreasing concomitant sevoflurane concentration. Fentanyl 4 mu g kg(-1) suppressed the changes in HR and MAP more effectively than fentanyl I or 2 mu g kg(-1) at sevoflurane concentrations close to MAC-TI.