In a double-blind, cross-over study in six healthy volunteers, the effects of different oral doses of midazolam (10, 20 and 40 mg), or 0.15 mg kg-1 midazolam administered intravenously and of placebo were investigated. Plasma concentrations of midazolam and of its active alpha-hydroxy metabolite were measured at the same time. The effect was assessed using objective and subjective methods (reaction time, tracing test, subjects' self-assessment and investigator's subjective assessment). The respective time courses of the plasma concentration and of the effect (reaction time, number of errors in the tracing test) were almost identical. Peak plasma levels and maximum effects were attained within 30 min. In general, the effect after intravenous injection of 0.15 mg kg-1 and after an oral dose of 10 mg midazolam lasted for 2 h following administration and its duration was doubled (i.e. to 4 h) after the 20 mg oral dose. Between the logarithm of the plasma concentration and the effect, there is a sigmoidal relationship that is virtually time independent. Particularly in the first few hours after oral administration the effect is intensified by the alpha-hydroxy metabolite of midazolam which is formed by first-pass metabolism. At identical plasma concentrations of midazolam, the oral dose produced more marked effects than did the intravenous administration. Correlation of the measured effects with the total (midazolam + alpha-hydroxy midazolam) plasma concentration reveals a closer sigmoidal relationship.