This prospective, phase II study was carried out to evaluate the efficacy and safety of midazolam as a sedative agent in nine critically ill Japanese patients without coma who had been admitted to an intensive care unit. An adequate level of sedation (sedative score [SS] 4-6) was induced by midazolam 0.14 - 0.39 mg/kg and maintained with a dose range of 0.08 - 0.24 mg/kg per h. One patient was excluded from the efficacy analysis since optimum sedation was not maintained by continuous infusion. The percentage of time during which SS 5 was achieved varied among the patients. The most frequent adverse events were hypotension, partial pressure of carbon dioxide elevation and premature ventricular contraction. Population pharmacokinetic analysis suggested that lower drug clearance rates were correlated with the presence of complications. The 50% of effective concentration values of SS 5, estimated by logistic regression analysis, varied greatly (mean, 251 ng/ml). In conclusion, midazolam infusion achieved successful sedation in this critical care setting. The optimum dose differed among the patients, however, and was influenced by the presence of complications.