The aim of this study was to estimate the pharmacokinetics ( PK) of arbekacin in burn patients using a population - PK approach. Therapeutic drug monitoring data consisting of 126 plasma concentrations ( including 17 values that were below the quantitation limit) from 47 burn patients were retrospectively analyzed using a mixed effect method ( NONMEM, ver. 6.0). Covariates, such as burn index, age, sex, among others, were tested on the basic one-compartment model. In the basic model, positive correlations of body weight ( WT) and creatinine clearance ( CLcr) on total clearance ( CL) and volume of distributions ( V) were assumed. In the final model, V increased with burn index ( BI). The final model was: CL(L/h) =3.18x WT/ 70+4.49x CLr/120; V(L) = 27.5x WT/70 + 0.28x (BI-23.5). Between- subject variability in terms of CL and V were 35 and 39%, respectively. The CL of our burn patients was significantly greater than that reported in unburned patients, and V increased proportionally with increasing BI.