To determine population pharmacokinetic parameters of vancomycin in neonates. This was a retrospective design, with prospective validation. Two hundred ten sequential neonates were evaluated at the neonatal intensive care units of Minneapolis Children's Medical Center and Children's Hospital of St. Paul. Five hundred twenty serum concentrations from 192 patients were included. A mean ± SD gestational age of 29.5 ± 5.1 weeks, postnatal age of 16.5 ± 19.6 days, and dosing weight of 1492 ± 1053 gm described the population. Thirty additional patients were studied for validation. Dosing, serum concentrations, and 28 covariates were collected. Data were evaluated with NONMEM. Forward selection and backward elimination regression identified significant covariates. One‐ and two‐compartment population pharmacokinetic parameters and predictive performance of the models were measured. Two‐compartment final regression equations were as follows: Clearance (CL) = 0.0590 L/kg/hr (multiplied by 0.460 if exposed to dopamine and 0.643 if gestational age was ≤32 weeks), central volume (Vc) = 0.440 L/kg, intercompartmental clearance (Q) = 0.0313 L/hr/kg, and steady‐state volume of distribution (Vss) = 0.764 L/kg. Interindividual variability was 40.6% for CL, 54.1% for Vss, and 16.8% for Vc. Residual variability was 3.3 µg/ml. One‐compartment final regression equations were: CL = 0.0626 L/kg/hr (multiplied by 0.455 if exposed to dopamine and 0.656 if gestational age was ≤32 weeks), and Vd = 0.496 L/kg. Differences in relative performance were insignificant by use of one‐ or two‐compartment parameters. Gestational age ≤32 weeks and concurrent use of dopamine were significant factors in prediction of vancomycin clearance, α Half‐lives of 2.8 to 3.7 hours and β half‐lives of 13.4 to 33.7 hours suggest that some individuals in this neonatal population have considerably longer half‐lives than those previously reported.