Theophylline is commonly used in neonatology for the treatment and prophylaxis of apnoea of prematurity, and during ventilator weaning. NONMEM was used to study the population pharmacokinetics of intravenous and oral theophylline from retrospective drug monitoring data in 82 premature neonates, weighing < 1500 g at birth, and 5 32 weeks gestational age.Clearance (CL), volume of distribution (V), and oral bioavailability (Fl) from liquid preparations were modelled alone, and under the influence of demographic and clinical covariates, assuming a 1-compartment model with first-order elimination.The final population models with influential co-variates were as follows: CL(1 h -I) = 0.0000123 *body weight (8) + 0.000377 *postnatal age (days); V (1) = 0.000937 *body weight (g); F=0.918. The CL was lower and V was higher than previously reported for less premature neonates, term babies, and older children. Predictive performance of the population models was evaluated by Bayesian forecasting in a similar, but independent cohort of 30 infants. There was statistically insignificant bias and imprecision between measured and predicted serum theophylline concentrations. Based on the validated population models, recommended maintenance theophylline dosages are provided for infants aged between 2 and 50 days, and weighing 700 to 2000 g.