During a 5-year period, 1997 to 2002, therapeutic drug monitoring of midazolam plasma concentrations in combination with the level of sedation as assessed by the Ramsay sedation scale was performed in 648 critically ill patients requiring artificial ventilation. In a subgroup of 189 patients sepsis-related organ failure assessment procedure was additionally performed. A total number of 3354 samples were analyzed. Significantly reduced clearance of midazolam was observed within the first 4 days of midazolam treatment of critically ill patients. As a result, accumulation of midazolam and its metabolites occurred within the first week of treatment. In contrast, parameters such as serum bilirubin or creatinine, which are commonly used to adapt drug therapy to organ dysfunction, showed significant changes with a delay of more than 10 days as compared with the findings of midazolam monitoring. Midazolam plasma concentrations showed a good correlation with the sedative capacity of the drug (r2 = 0.906). However, a great variability of the drug effect between patients could be demonstrated, which, as a consequence, may complicate the development of dosing strategies based on midazolam plasma concentrations to better control sedation in critically ill patients. Furthermore, patient age seems to be an important factor for the considerable variability of the sedative effect of midazolam. To achieve a certain levels of sedation, significantly lower midazolam infusion rates as well as plasma concentrations were required as the patients age increased. No significant sex-related differences could be observed for any pharmacologic parameter obtained in this study. Our findings suggest that midazolam therapeutic drug monitoring might be a useful tool to individualize midazolam therapy, especially in critically ill patients developing organ dysfunction and requiring long-term sedation to minimize the risk of drug accumulation and excessive sedation.