Objective: To describe the pharmacokinetics of midazolam, a water-soluble benzodiazepine with a short half-life, in critically ill neonates. Hypothesis: Midazolam clearance is reduced in neonates compared with clearance in children, and the doses currently in use, which are derived from pediatric studies, are excessive. Patients and Methods: This population study was conducted in 187 neonates requiring intravenous sedation for artificial ventilation. The 531 midazolam concentration measurements obtained were analyzed by use of NONMEM and a two-compartment model with four parameters: clearance (CL), central volume (V,), peripheral volume (Vp), and intercompartmental clearance (Q). The influence of birth weight (range, 700 to 5200 gm), gestational age (range, 26 to 42 weeks), postnatal age (range, 0 to 10 days), and comedications were investigated.Results: CL and Vc (mean ± SE) were found to be directly proportional to birth weight (CL = 0.070 -± 0.013 L/kg/hr; Vc = 0.591 ± 0.065 L/kg). The CL was 1.6 times higher in neonates with a gestational age of more than 39 weeks. It was 0.7 times lower in neonates receiving inotropic support.The postnatal age had no apparent effect on midazolam kinetics. The VP and Q (mean ± SE; 0.42 0.11 L and 0.29 ± 0.08 L/hr, respectively) were not influenced by any of the covariates studied. There was a large interindividual variability for the pharmacokinetic parameters. Conclusion: The mean midazolam doses required for critically ill neonates are lower than those required for older infants. (CLIN PHARMACOL THER 1994;56:615-25.)