SummaryPopulation pharmacokinetic parameters of gentamicin in preterm neonates on a once-daily dosage regimen of 3.0 mg/kg given intravenously every 24 hours were established prospectively. In 34 preterm neonates with a mean gestational age of 32 ± 4 (SD), 182 serum gentamicin levels (91 peak/trough pairs) were determined. Individual adjustments of dose or dosage interval were calculated by computer-aided Bayesian forecasting. The parametersVd, ke, and CL for each patient were obtained by the nonparametric estimation of maximization method. The predictive power of the model was calculated and the pharmacokinetic estimates were statistically analyzed with SPSS/PC. Cluster analysis showed a division into 2 subpopulations (designated 1 and 2) on the basis of postnatal age. The mean ± SD postnatal age of subpopulation 1 (n = 29) was 6 ± 2 days (range 1–7) and of subpopulation 2 (n = 5) 15 ± 4 days (range 12–24). The mean ± SD gentamicin relative clearances of subpopulation 1 and subpopulation 2 were 0.0515 ± 0.0128 and 0.1026 ± 0.0102 L kg−1hr−1, respectively (p < 0.05). The mean ± SD values forVd(Lkg−1) in both populations 1 and 2 were 0.6916 ± 0.1670 and 0.7509 ± 0.1961, respectively (not significantly different). For kethese data were 0.0744 ± 0.0200 and 0.1366 ± 0.0522 (p < 0.05). Statistics showed that the data forVdand keof subpopulation 1 were normally distributed (Vdand keskewness 1.61 and 1.46; kurtosis 3.09 and 3.10 respectively). The model yielded a bias of −0.11 mg/L and a precision of 0.36 mg/L. It is recommended that gentamicin be started in a dosage of 3.5 mg/kg intravenously once-daily under close monitoring.