Cerebral blood flow (CBF) and cerebral metabolic rate for oxygen (CMRO2) were measured by a modification of the Kety-Schmidt technique using i.v. xenon-133 in 20 patients undergoing craniotomy for supratentorial cerebral tumours. Anaesthesia was induced and maintained with midazolam, fentanyl and nitrous oxide. Pancuronium was given for neuromuscular block. The lungs were ventilated to normocapnia. The first flow measurements were performed approximately 1 h after induction of anaesthesia. At the end of operation the patients were allocated to two groups. Ten patients were given flumazenil 0.01 mg kg-1 and 5 min later the second flow measurement was performed. In the other 10 patients the second flow measurement was performed before the administration of flumazenil. Plasma concentrations of midazolam were measured at the time of each measurement of CBF. There was no difference between the groups in plasma concentration of midazolam, CBF or CMRO2. Flumazenil had no effect on CBF and CMRO2.