The new benzodiazepine antagonist flumazenil represents another approach to the ever-present problem of recurring respiratory depression after anesthesia with flunitrazepam and fentanyl. Objective and subjective side effects of flumazenil were studied in comparison with the opiate antagonists naloxone and nalbuphine. METHODS. One hundred fifty surgical patients, ASA I or II, aged 18-65 years were studied. After premedication with atropine 0.5 mg and flunitrazepam 0.5 mg anesthesia was induced with flunitrazepam 0.5 mg, fentanyl 0.1 mg and etomidate 10 mg and maintained with N2O/O2 2:1 and additional increments of 0.1 mg fentanyl as required. Relaxation for intubation and surgery was obtained with non depolarizing muscle relaxants. After the operation the patients were extubated and then flumazenil 0.4 mg, naloxone 0.05 mg, or nalbuphine 20 mg was given i.v. (randomized and double-blind). In 15 patients blood pressure and heart rate were monitored. In all patients postoperative pain was assessed by the time interval between administration of the antagonist and need for the first analgesic medication. On the 1st postoperative day recall of postoperative events and of pictures shown 5, 30, 60, 120, and 240 min after administration of the antagonist was tested. The patients were interviewed a second time for side effects on day 3-6 after the operation. RESULTS. The three antagonists produced no significant effects on arterial pressure and heart rate. There were no differences between the antagonists in the incidence of postoperative nausea and/or vomiting or postoperative pain. After flumazenil, a significant transient increase in vigilance and better recall of postoperative events was noted within 5 and 30 min after administration of the drug. CONCLUSION. On the basis of the objective clinical findings, there is no reason to prefer either benzodiazepine or opiate antagonists after flunitrazepam and fentanyl. However, postoperative amnesia can be reduced by flumazenil if this is desirable.