Gentamicin in combination with benzylpenicillin, is widely used as empiric treatment in early onset neonatal sepsis. The aim of this study is to explore influence of potential covariates on the pharmacokinetic parameters of gentamicin in very premature neonates. A population pharmacokinetic model was developed using NONMEM V (version 1.1). Serum gentamicin concentrations retrospectively collected from routine therapeutic drug monitoring database from neonates with a gestational age (GA) of less than 32 weeks were analyzed using a one-compartment open model with first order elimination. The stability of the model developed was evaluated using non-parametric bootstrapping with replacement. A total of 312 serum gentamicin levels were obtained from 139 preterm neonates with a mean gestational age of 28.6 ±1.9 weeks. The typical population parameters estimated for clearance and volume of distribution for the population (28.8 weeks of post conceptional age and body weight of 1.15 kg) were 0.0351 Lh-1 and 0.545 L, respectively. The clearance observed in this study was lower than those previously reported. The between-subject variability (CV%) for clearance was 24.9% while the variability for volume of distribution was highly co-related with that of clearance. In the combined residual unexplained variability model, the proportional error component (CV%) was 29.4%, the additive error component (SD) was 0.26 mgL-1. The results revealed that the most important predictors for gentamicin clearance are size (weight) and post-conceptional age while weight alone significantly influenced the volume of distribution. The diagnostics for model performance showed the model to be unbiased. Dosing guidelines incorporating these factors were developed using simulations from the model parameters for clinical setting.