The pharmacokinetic parameters of vancomycin in a neonatal population have been characterized to enable development of optimum dosage guidelines for neonatal intensive-care units and to examine the relationship between these pharmacokinetic parameters and various demographic, developmental and clinical factors which might be associated with changes in the kinetic profile of vancomycin. Forty-four infants (twenty-five males and nineteen females) with suspected or proven Gram-positive infection and who received intravenous vancomycin between October 1993 and December 1996 were included in this retrospective analysis. Gestational age ranged from 25 to 40 weeks and postconceptional age at the time of the study ranged from 28 to 45 weeks. Sixty case-studies were obtained from the forty-four patients, with one period of study corresponding to one week or one cycle of therapy. Vancomycin pharmacokinetic parameters were determined by use of a one-compartment model. By regression analysis the current weight (g) was shown to be the stronger covariate, and both vancomycin clearance (L h(-1)) and volume of distribution (L) had to be normalized. The vancomycin volume of distribution depended on the postconceptional age with a cut-off at 32 weeks, whereas vancomycin clearance depended on the presence or absence of concomitant treatment with indomethacin or of mechanical ventilation, or both. On the basis of the pharmacokinetic parameters obtained we suggest initial dosage guidelines for vancomycin ranging from 10 mg kg(-1) every 8 h to 10 mg kg(-1) every 12 h, depending on the demographic and clinical characteristics of the patients. The results obtained enabled application of better a priori and a posteriori dosage schedules to infants in neonatal intensive-care units by use of the Bayesian approach, although further prospective study is recommended before direct extrapolation to patients in other settings.