A randomized multicenter study in intensive care unit (ICU) patients, evaluated the capacity of a Bayesian method to obtain an optimal first isepamicin (ISP) peak of 80 mg/L in comparison to a fixed loading dose (LD). Patients (n=236) over 18 years of age were enrolled from 6 September 1997 to 17 July 1999 and randomly assigned to received ISP in a calculated dose (CD) or a loading dose (LD) of 25 mg/kg body weight. The CD was estimated using a specific population model with Bayesian methodology implemented in the PKS program (Abbott PKS, Abbott Diagnostics, Rungis, France). The data required included age, body weight, height, gender and serum creatinine. ISP disposition is described by a one-compartment model. Blood samples were drawn 1 and 24 h after the start of infusion for fluorescence polarization immunoassay measurement of serum ISP concentrations. The predictive performance was assessed by computing bias and precision. Peak concentrations were significantly higher in CD group than the LD group (84.2 +/- 28.6 vs. 74.7 +/- 24.1 mg/L, respectively; P=0.008), but trough levels were comparable. The optimal ISP peak was attained by a significantly higher percentage of CD patients (P=0.018), and by significantly more CD patients on mechanical ventilation (P=0.025), and with simplified acute physiological scores (SAPS) > 35 (P=0.002). Pharmacokinetic parameters were similar for the two groups with large interindividual variations. Mean (+/- SD) volume of distribution of ventilated patients (72%) was significantly higher than of nonventilated patients (23.31 +/- 7.35 vs. 20.60 +/- 6.30 L, respectively; P=0.001). No relationship was found between the volume of distribution and SAPS. Total clearance was significantly correlated with estimated CLCR (creatinine clearance) (P=0.0001). Precision (RMSE) is better for CD than for LD strategy, respectively 27.96 and 28.66 mg/L. The Bayesian method was significantly more accurate and performed particularly well in ventilated patients and patients with high SAPS, compare to an LD of 25 mg/kg to obtain a first ISP peak of 80 mg/L in ICU patients. Therefore, a fixed dose of 28.5 mg/kg would be also adequate to reach a peak of 80 mg/L.