Two methods are commonly used to analyze the pharmacokinetics of ABK : the Sawchuk-Zaske (SZ) method, which uses a one-compartment model, and the Bayesian (BS) method, which uses population parameters based on a two-compartment model. Since it seemed possible that serum ABK concentrations predicted from pharmacokinetic parameters calculated by these two methods might differ, we carried out a comparison. We performed TDM after administering ABK to 18 patients with pneumonia and evaluated the peak and trough concentrations predicted by the two methods. The predicted between peak concentrations for the SZ method and BS method were well correlated and so were the predicted between trough concentrations for both methods. However, the peak and trough concentrations predicted by the BS method were higher than those for the SZ method by 3.9± 1.5 and 1.1± 0.8μg/mL, respectively, which suggests that actual peak concentrations could be lower than the therapeutic range if the target peak levels are set between 7 and 11μg/ mL when designing the dosage regimen by the BS method. We also compared measured values with values predicted based on patients' pharmacokinetic parameters (population means) using the initial dosage design function of TDM software for Habekacin ® (Research Committee for Anti-MRSA Drugs). The relationship between the measured values (y) and the predicted values (x) was represented by the regression equation y=1.075x -0.309 (R²= 0.729, n= 36), demonstrating the usefulness of the initial dosage design.