The effect of many intravenous sedative and hypnotic drugs on subcortical and early cortical evoked responses has been investigated by many authors. Because of the great inter- and intraindividual variabilities late cortical evoked components (EP) are thought to be unable to reflect the degree of anesthetic depth. However the impact of centrally acting anesthetic drugs on EP has not been thoroughly studied. The present study investigates the influence of midazolam and the specific benzodiazepine antagonist flumazenil on acoustical (AEP) and somatosensory (SEP) evoked cortical responses. In addition the following parameters were recorded: heart frequency, mean arterial blood pressure, arterial oxygen saturation and in approximation the end-expiratory carbon dioxide tension. In 12 healthy volunteers (age: 25-36 y) the following combinations of drugs were given in several sessions: Midazolam-flumazenil, midazolam-ringer's solution, ringer's solution-flumazenil. In 6 subjects the study was performed according to a double blind placebo-controlled experimental protocol. Midazolam led to a suppression and a prolongation of latencies of components: N1/P2; P2/N2 (AEP); P100/N150; N150/P220 (SEP). Flumazenil (0.3 mg) led to a prompt restoration of all cortical components with however still reduced amplitudes compared to control. At the end of the observation period, 50 min after application of the second drug, EP-components did not differ significantly between the groups given midazolam-ringer's solution and midazolam-flumazenil. Flumazenil given alone had no significant effect on the AEP- and SEP-recordings, so no intrinsic activity of flumazenil could be substantiated.