Population pharmacokinetic (PK) modeling and Monte Carlo simulation (MCS) were used to describe the pharmacodynamic profile of cefepime in the both plasma and cerebrospinal fluid (CSF). Plasma and CSF cefepime data were obtained from a PK study of 7 hospitalized patients with external ventricular drains. Concentration-time profiles in plasma and CSF were modeled using a 3-compartment model with zero-order infusion and first-order elimination and transfer. Estimates of the PK parameters were identified in the Big Non Parametric Adaptive Grid with adaptive gamma (BigNPAG) program of Leary, Jelliffe, Schumitzky, and Van Guilder. MCS (10,000 subjects) was performed to estimate the probability of attaining the targets of free plasma concentration (20% protein binding) and total drug CSF concentration of 50-100% T>minimal inhibitory concentration (MIC) for MICs 0.06-8 mg/L for cefepime 2 g, iv, every 8 h (0.5-h infusion); cefepime 2 g, iv, every 12 h (0.5-h infusion); and cefepime 2 g (0.5-h infusion) once and 250 mg/h continuous infusion. After the Bayesian step, the observed-predicted regression and r(2) for plasma and CNS were as follows: plasma, observed=0.984 x predicted+2.570, r(2)=0.944; CSF, observed=0.785 x predicted+0.868, r(2)=0.821. The median penetration of cefepime as measured by AUC(CSF)/AUC(plasma) was 7.8%. In the MCS, the target attainment rates in plasma for 60-70% fT>MIC were high at each MIC value between 0.03 and 8 microg/mL for each regimen examined. In CSF, none of the regimens achieved 50-100% T>MIC for>80% of patients for MICs>0.5 mg/L.