The aim of this study is to determine the pharmacokinetic parameters of midazolam (MDZ) and its active metabolites, namely 1-hydroxymidazolam (1-OH-MDZ) and 1-hydroxymidazolam glucuronide conjugate (1-OH-MDZG) in patients with various degrees of severity of sepsis syndrome. A 6-compartment disposition model with combined intermittent oral administration and continuous intravenous infusion with first-order elimination was used. The model was fitted to serum MDZ concentration data (473 observations), serum 1-OH-MDZ concentration data (474 observations), serum 1-OH-MDZG concentration data (470 observations), and urine 1-OH-MDZG concentration data (883 observations) obtained prospectively from 55 critically ill adults receiving continuous infusions of MDZ (one received oral administration prior to intravenous administration) for the induction and maintenance of sedation. All serum and urine concentrations of MDZ, 1-OH-MDZ and 1-OH-MDZG were estimated by a locally developed liquid chromatography-mass spectrometry assay method. Population parameters were estimated with NONMEM. The predictive performance of the model was evaluated by plots of visual predictive checks using 200 simulated datasets derived from the final model. The volume of distribution for both central and peripheral compartments, were affected by body weight (which reflects the severity of fluid retention). The clearance of serum MDZ to other sites was influenced by the hematologic component of the Sepsis-induced Organ Failure Assessment (SOFA) score while the clearance of serum and the clearance of serum 1-OH-MDZG by both the renal component of the SOFA score and noradrenaline administration. A 41% reduction in the clearance of MDZ to 1-OH-MDZ was associated with noradrenaline administration. The between-subject variability of the pharmacokinetic parameters remained high; from 62% in clearance of serum MDZ to 1-OH-MDZ to 197% in clearance of serum 1-OH-MDZG. Typical population parameter estimates of clearance for a 70-kg patient were: MDZ to other sites/routes, 6.7 L/h with normal coagulation status, and reduced to 0.015 L/h in severe coagulapathy; MDZ to 1-OH-MDZ, 3.9 L/h and reduced to 2.3 L/h with noradrenaline administration; 1-OH-MDZ, 16.8 L/h and 1-OH-MDZG, 18. 5 L/h in normal renal function and not requiring noradrenaline support to 1.1 L/h in severe renal impairment and requiring noradrenaline administration; glucuronide fraction of 1-OH-MDZ clearance is 0.59 and the renal fraction of 1-OH-MDZG clearance is 1.24. The central and peripheral distribution volumes were 13.8 L and 71.8 L, respectively; while the inter-compartmental clearance was 35.4 L/h. The bioavailability of MDZ was fixed at 0.8 according to previously reported data and the absorption rate constant, 0.97 /h. Model evaluation by visual predictive checks demonstrated that there was no appreciable bias in the prediction. In conclusion, the pharmacokinetic profile of MDZ and its active metabolites in critically ill patients receiving continuous intravenous MDZ infusion was adequately estimated by a 6-compartment model. Important predictive covariates for MDZ infusion rate were body weight, the hematologic and renal components of the SOFA score as well as administration of noradrenaline.