In a double-blind placebo-controlled prospective clinical trial we studied the efficacy and safety of the benzodiazepine antagonist, flumazenil. In 23 patients admitted to the Intensive Care Unit with coma due to overdose with benzodiazepines or other sedatives, flumazenil i.v. (up to 2 mg or placebo) was given. In 13 patients given flumazenil the Glasgow Coma Scale (GCS) increased significantly from 4.9 to 7.8 (p less than 0.05). Six of these 13 patients, including mainly benzodiazepine mono-intoxications, needed only one series of injections (up to 1.0 mg flumazenil); the GCS increased thereby from 4.5 to 10.7 within a maximum of 5 min (p less than 0.01). In the remaining 7 patients, needing two series of injections of flumazenil (up to 2.0 mg), GCS did not rise significantly and coma was related to intoxications with nonbenzodiazepine sedatives, flunitrazepam and in one patient, encephalitis. In the 10 patients receiving placebo, the GCS did not change. A significant increase in the GCS from 5.5 to 10.8 (p less than 0.001) was, however, observed when flumazenil (up to 1.0 mg) was given after placebo. In patients with EEG monitoring the changes in waveform pattern paralleled the clinical response. Effects could be detected within 1-2 min after flumazenil injection and lasted up to 45 min. There were no adverse reactions or benzodiazepine withdrawal symptoms. We conclude that flumazenil is an effective and safe drug in the treatment of benzodiazepine overdose. The use of flumazenil is of diagnostic value in mixed-drug intoxications or coma of unknown origin and is of therapeutic importance for reversal of benzodiazepine intoxications.