1. Midazolam, a short-acting benzodiazepine, has been considered a probe for estimating hepatic and intestinal cytochrome P450 (CYP) 3A activity in humans. The aim of the present study was to evaluate the pharmacokinetics and pharmacodynamics of midazolam administered intravenously (i.v.) and orally (p.o.) at relatively low doses to healthy volunteers. 2. The present study was an open-label, single-sequence trial in three phases distinguished by differing doses of midazolam. Plasma concentrations of midazolam and its metabolites, as well as pharmacodynamic parameters, were measured simultaneously after administration of 5, 15 and 30 microg/kg, i.v., midazolam and 15, 50 and 100 microg/kg, p.o., midazolam. 3. The area under the concentration-time curve (AUC) of midazolam was significantly correlated with dose after both i.v. and oral administration (both P < 0.001). The AUC(0-6) of midazolam after oral administration was also well correlated with the area under the effect curve for peak saccadic velocity (PSV; P < 0.018), postural sway area (PSA; P < 0.069) and mental sedation as measured on a visual analogue scale (VAS; P < 0.054), but not for critical flicker fusion. 4. The present study has shown that the pharmacokinetics of midazolam at relatively low doses are linear for both intravenous and oral dosing regimens. In addition, PSV, PSA and VAS may be useful for the simultaneous evaluation of the pharmacokinetics and pharmacodynamics of midazolam at subtherapeutic doses.