1 This study aimed not only to compare the pharmacokinetics of oral and intravenous doses of the new water-soluble benzodiazepine, midazolam, but also to study the effects on haemodynamics, sensorium, and memory performance. 2 Eight normal human volunteers each received a single 15 mg dose of midazolam base orally and intravenously in randomized sequence 2 weeks apart. Serial venous samples were obtained for 12 h after dosing. Vital signs, sensorium testing and memory testing using word lists were also performed. Computerized non-linear least squares curve-fitting of the two-compartment open model to the oral and intravenous data simultaneously yielded the following estimates: VI, 0.331 kg-1, Vdss, 1.081 kg-1, t2,x, 0.10 h, t½2,z, 1.89 h, ka 1.17 h-1 and bioavailability, 49%. The intravenous dose decreased the systolic pressure 22 mm Hg during the first half-hour and the oral dose had 50% less effect. Most subjects became drowsy halfway through the infusion and were only rousable to voice by its end. The sensorium was clear by 2-3 h. After oral dosing the peak sensorium effects of ataxiadysarthria were seen at 30 min and had cleared by 2 h. Memory testing showed that memory acquisition was markedly impaired for at least 90 min after the intravenous dose and slight recovery was apparent at this time after the oral dose. Memory performance was proportionately more impaired than the sensorium score. 3 We conclude that: (a) midazolam kinetics are characterized by rapid absorption, but incomplete bioavailability and rapid elimination, (b) midazolam intravenously may lower blood pressure significantly, and (c) the level of consciousness correlates poorly with the degree of memory impairment.