The role of protein binding in relation to the pharmacodynamics of alfentanil was investigated in 15 female and 13 male patients, aged 21-85 yr, ASA physical status 1 or 2, undergoing upper abdominal surgery. All patients had normal cardiac, hepatic, renal, and pulmonary function. None was receiving medication or had a history of alcohol or other drug abuse. Anesthesia was induced and maintained with 66% nitrous oxide in oxygen and alfentanil. Alfentanil was administered by a computer-controlled infusion pump. If, during surgery, the patient exhibited signs of inadequate anesthesia (i.e., response), the target alfentanil plasma concentration was increased by 50-100 ng/ml. If there was no response during a 15-min period, the target concentration was decreased by 50-100 ng/ml. Arterial blood samples were taken before any change of the target concentration and 4 min after the computer had indicated that the new target concentration had been reached. In addition, blood samples were taken before intubation, skin incision, and in the patients in whom ventilation recovered spontaneously before extubation. In the remaining patients a blood sample was taken before the administration of naloxone. Plasma alfentanil concentrations were determined by capillary gas chromatography. Alfentanil protein binding was determined by equilibrium dialysis in an arterial blood sample taken before induction of anesthesia. Alfentanil concentration-effect data were evaluated by logistic regression, where effect was either response or no response to perioperative stimuli. The average free fraction of alfentanil was 9.3 +/- 3.9% (range 3.7-19.1%). For intubation, skin incision, and postanesthesia ventilation, it was not possible to characterize the concentration-effect curves based on total plasma concentrations with logistic regression. However, adequate logistic regression characterization for intubation and skin incision could be obtained using the unbound concentrations. For the intraabdominal component of surgery the average total plasma concentration for which the probability of no response during surgery is 50% (Cp50 total) was 476 +/- 160 ng/ml (range 134-852 ng/ml). There was a significant negative correlation between Cp50 total and free fraction (r = -0.67, P < 0.001). There was no correlation between either Cp50 total or free fraction and age. The results of this study indicate that 45% (r2 = 0.45) of the observed variability in the Cp50 total of alfentanil could be explained by variability in the protein binding of alfentanil.