The aim of this study was to develop a population-pharmacokinetic model of ceftazidime in intensive care unit patients to include the influence of patients' characteristics on the pharmacokinetics. Forty-nine patients for model building and 23 patients for validation were included in a randomized study. They received ceftazidime at 2 g three times a day or as 6 g per day continuously. A NONMEM pharmacokinetic model was constructed, and the influences of covariates were studied. The model was validated by a comparison of the predicted and observed concentrations. A final model was elaborated from the whole population. Total clearance (CL) was significantly correlated with the glomerular filtration rate (GFR) calculated by modification of the diet in renal disease (MDRD), the central volume of distribution (V1) with intubation, and the peripheral volume of distribution (V2) with the reason for admission. The mean pharmacokinetic parameters were as follows: CL, 5.48 liters/h, 40%; V1, 10.48 liters, 34%; V2, 32.12 liters, 59%; total volume, 42.60 liters, 45%; and intercompartmental clearance, 16.19 liters/h, 42%. In the polytrauma population (mechanically ventilated), the time above the MIC at steady state never corresponds to 100% for discontinuous administration, and the target concentration of five times the MIC was reached with a 6-g/day dose only for patients with an MDRD of <150 ml/min. We showed that the GFR-MDRD, mechanical ventilation, and the reason for admission may influence the achieved concentrations of ceftazidime. Our model allows the a priori dosing to be adjusted to the individual patient.