To characterize the plasma concentration-effect relationship of flumazenil in the presence of a predefined midazolam level, a double-blind, placebo-controlled, randomized two-way crossover study was conducted in nine healthy male subjects. After reaching a criterion level of midazolam-induced depression of the Digit Symbol Substitution Test (DSST), volunteers received a dose of flumazenil (1.0 mg) or placebo over 1 minute, with the infusion of midazolam continued. Blood samples were collected, simultaneously with the DSST assessment, at predetermined intervals and were assayed for flumazenil and/or midazolam plasma concentrations. Pharmacokinetic-pharmacodynamic modeling techniques were used to estimate the equilibration rate constant (keo) between plasma concentration and effect for flumazenil; a sigmoidal maximum-effect model was used to relate the DSST score to the flumazenil plasma concentration. Flumazenil exhibited a rapid onset (the half-life of equilibration between drug concentration in the blood and drug effect was 3.3 minutes) and short duration of action (the flumazenil plasma concentration causing half-maximal effect was 7.4 ng/ml, which was reached about 1 hour after dosing). The results of this study also show the competitive nature of flumazenil as a midazolam antagonist.