The population pharmacokinetics of dapsone were examined in human immunodeficiency virus-infected patients receiving dapsone at a dosage of 100 mg twice weekly for the prevention of Pneumocystis carinii pneumonia. Nonlinear mixed-effect modeling was used to determine the best pharmacostatistical model for the data. A one-compartment open model with first-order absorption and elimination was used as the structural pharmacokinetic model. Several covariates were tested for their influence on pharmacokinetic parameters. Rifampin was found to increase the values of clearance/bioavailability (CL/F) and volume of distribution/ bioavailability (V/F) by approximately 70%. CL/F and V/F were 1.83 liters/h and 69.6 liters, respectively, for patients not taking rifampin. The effect of rifampin on the pharmacokinetic parameters of dapsone was appreciably less than expected on the basis of studies with healthy volunteers. Increased bilirubin levels were associated with a significant decrease in the absorption rate constant (Ka). However, this finding may be considered clinically irrelevant because the post hoc Bayesian estimates of Ka for patients with high bilirubin levels ( > 1.2 mg/dl) were at the lower bound of the values for patients with normal bilirubin levels. The value of Ka was 0.957 h-1 for a patient with a bilirubin level of 0.7 mg/dl. After inclusion of covariates in the model, the interpatient variability was 35% for CL/F, not significant for V/F, and 85% for Ka. Simulation of plasma concentration-versus-time curves indicated that the administration of 100 mg of dapsone biweekly is associated with sustained dapsone levels in the plasma of the majority of the patients. Dosage adjustments for patients concomitantly treated with rifampin may be necessary.