Background: Vancomycin is an effective glycopeptide antibiotic against gram-positive infections. Vancomycin has a relatively narrow therapeutic window (5~15 mg/L) and has potentially severe adverse effects such as nephrotoxicity. In order to maintain effective drug concentrations and reduce the possibility of toxicity, therapeutic drug monitoring is useful. The pharmacokinetics of vancomycin is known to differ between pediatric and adult patients because of difference in body size, renal function, etc. This study aimed to describe the population pharmacokinetics of vancomycin in Korean pediatric patients and to evaluate the predictive performance of the population pharmacokinetic model. Methods: Vancomycin dosage histories, sampling times, concentrations and demographic data were collected retrospectively from therapeutic drug monitoring reports. Patients were randomly allocated to either the index data set (n=92) or a validation data set (n=25). The index data set was comprised of 159 concentration measurements. The population pharmacokinetic parameters of vancomycin were estimated by nonlinear mixed-effect modeling using first-order conditional estimation with interaction method. The influence of covariates such as age, weight, height, gender, serum creatinine, creatinine clearance and concomitant medicine (aminoglycosides) were assessed. The predictive performance of the models was evaluated using the validation data set. Bias and precision were assessed by mean prediction error and root mean squared error. Results: The two compartment model was adopted and the final population pharmacokinetic model related clearance to weight and serum creatinine, and central volume of distribution to weight. The final population pharmacokinetic parameters were modeled as follows: clearance (L/h) = $0.0722;{ imes};weigth^{1.3};(kg);{ imes}$ (1 - serum creatinine (mg/dL) ${ imes};0.506$), central volume of distribution (L) = $0.0709;{ imes};weight^{1.61};(kg)$, $k_{12};(1/h);=;0.288$, $k_{21};(1/h);=;0.14$. The coefficient of variations (CVs) of interindividual variability in clearance and central volume of distribution were 25.5% and 18.6%, respectively. The residual variability was 23.5%. Predictive performance of this model was evaluated using the validation data set. The mean prediction error and root mean squared prediction error were -0.17 and 2.10 mg/L, respectively. Conclusions: A population pharmacokinetic models of vancomycin in Korean pediatric patients was newly developed. Weight and serum creatinine were significant factors affecting vancomycin disposition. The predictive performance of this model was reasonably acceptable. The population pharmacokinetic parameter values of Korean pediatric patients obtained from this model may be used to optimize vancomycin therapy.