Abstract: OBJECTIVE To design a individualized dosage regimen depend on the individual pharmacokinetic parameters which was calculated using the Byes feedback model and the constructed population pharmacokinetic(PPK) modle with the single point trough concentration of vancomycin. And the vancomycin PPK were established by using the sparse TDM datas from the adult patients treated with vancomycin in the 3 hospital of grace medical center. METHODS In the present paper, the concentration of 235 serum samples collected from 128 vancomycin treated patients were determined. One compartment pharmacokinetic model of intravenous injection integrated in Kinetica software was used for fitting the basic model by the calculating modle of maximum likelihood(EM) and Bayesian(Bayes). Forward stepwise regression was used to study the relationships between the pharmacokinetic parameters (elimination rate constant, Kel; the apparent volume of distribution, Vd) and the covariants (serum creatinine, Scr; Age; Weight, Wt; Sex; Drug combination) and then fitted the final modle given the the fixed effect model by EM and Bayes again. In the end, the model was evaluated using the method of internal bootstrap and external validation method. RESULTS The final PPK modle in the current study was Kel=θ₁×(Scr)^θ₂×(Sex)^θ₃, V=θ₄×(Scr)^θ₅×(Age)^θ₆×(Wt)^θ₇(θ₁=0.18; θ₂=-0.19; θ₃=-0.31; θ₄=20.85; θ₅=-0.52; θ₆=-0.23; θ₇=0.98). The population pharmacokinetic parameters (Cl and Vd) according to the final model were 5.0 L·h^-1 and 66.9 L, the Bayes predict error were 0.8 L·h^-1 and 9 L. CONCLUSION The established model of vancomycin PKK model can well reflect the population pharmacokinetic characteristics of Chinese adult vancomycin treated patients with low predicted error of the single point Byes feedback and provide the theoretical and practical basis for improving treating efficacy and reducing adverse reactions.