Objective: Midazolam is a water-soluble benzodiazepine. Flumazenil is a potent antagonist of midazolam-induced sedation. Physostigmine has also been shown to reverse benzodiazepine sedation in anecdotal reports. The aim of this study was to quantitatively characterize the reversal of midazolam- induced changes in electroencephalogram (EEG) by physostigmine compared to flumazenil and placebo. Methods: Twelve healthy male subjects received 5 mg midazolam as an intravenous infusion over 15 minutes. Fifteen minutes after the end of infusion, single doses of either 0.4 mg flumazenil, 0.5 mg physostigmine, or placebo (physiologic saline solution) were administered as intravenous injections in a randomized crossover fashion. Midazolam serum concentrations were measured using liquid chromatography-tandem mass spectrometry. The time from the start of injection until awakening was noted and the EEG was measured. Results: Four subjects were excluded from further pharmacokinetic and pharmacodynamic analysis because no midazolam-induced changes on EEG alpha power could be observed in each of the three study periods. The pharmacokinetics of midazolam were not influenced by flumazenil or physostigmine. Midazolam induced a decrease in EEG alpha power (7.50 to 11.25 Hz) compared with baseline (P < .05). After injection of flumazenil and physostigmine, an increase in EEG alpha power was observed, whereas placebo did not affect alpha power. Subjects opened their eyes 25.2 ± 1.1 minutes after the placebo injection was begun, whereas subjects awoke after 6.2 ± 2.7 minutes and 15.4 ± 3.4 minutes after they received flumazenil and physostigmine, respectively (mean ± SEM; P < .001). Conclusion: Physostigmine and flumazenil antagonized midazolam-induced sedation. This suggests that a reversible central anticholinergic mechanism may be involved in the sedative action of midazolam.