The pharmacokinetic variability of voriconazole (VCZ) in immunocompromised children is high, and adequate exposure, particularly in the first days of therapy, is uncertain. A population pharmacokinetic model was developed to explore VCZ exposure in plasma after alternative dosing regimens. Concentration data were obtained from a pediatric phase II study. Nonlinear mixed effects modeling was used to develop the model. Monte Carlo simulations were performed to test an array of three-times-daily (TID) intravenous dosing regimens in children 2 to 12 years of age. A two-compartment model with first-order absorption, nonlinear Michaelis-Menten elimination, and allometric scaling best described the data (maximal kinetic velocity for nonlinear Michaelis-Menten clearance [] = 51.5 mg/h/70 kg, central volume of distribution [] = 228 liters/70 kg, intercompartmental clearance [] = 21.9 liters/h/70 kg, peripheral volume of distribution [] = 1,430 liters/70 kg, bioavailability [] = 59.4%, = fixed value of 1.15 mg/liter, absorption rate constant = fixed value of 1.19 h). Interindividual variabilities for , , , and were 63.6%, 45.4%, 67%, and 1.34% on a logit scale, respectively, and residual variability was 37.8% (proportional error) and 0.0049 mg/liter (additive error). Monte Carlo simulations of a regimen of 9 mg/kg of body weight TID simulated for 24, 48, and 72 h followed by 8 mg/kg two times daily (BID) resulted in improved early target attainment relative to that with the currently recommended BID dosing regimen but no increased rate of accumulation thereafter. Pharmacokinetic modeling suggests that intravenous TID dosing at 9 mg/kg per dose for up to 3 days may result in a substantially higher percentage of children 2 to 12 years of age with adequate exposure to VCZ early during treatment. Before implementation of this regimen in patients, however, validation of exposure, safety, and tolerability in a carefully designed clinical trial would be needed.