The dose proportionality and bioequivalence of the capsule formulations used in clinical trials and the proposed commercial formulations of stavudine were assessed in an open-label, single-dose, randomized four-way crossover study in 16 asymptomatic HIV-infected males. One capsule of stavudine (5, 10, 20, or 40 mg) was administered orally to each subject in each of the four treatment periods. Serial blood samples were collected for 10 h after each dose and the plasma was assayed for intact stavudine by a validated radioimmunoassay method. The plasma concentration-time data were subjected to noncompartmental pharmacokinetic analysis. For doses ranging from 5 to 40 mg, mean C-max and AUC(0-infinity), values were in the range of 110.36-889.34 ng mL(-1) and 246-46-1945.97 h ng mL(-1) respectively. The mean C-max and AUC(0-infinity)f stavudine increased in a dose-proportional manner. Irrespective of the dose, mean C-max values were observed at a median t(max) of 0.75 h or less. Mean tip values were 1.97, 1.77, 1.67 and 1.66 h for the 5, 10, 20, and 40 mg capsules, respectively. For bioequivalence assessment, C-max and AUC(0-infinity), values were normalized to the 10 mg dose since these parameters were dose proportional. The 10 mg capsule formulation used in phase-3 clinical trials was chosen as the reference. The relative bioavailability estimates and 90% confidence limits for the dose-normalized C-max values with the 10 mg capsule as the reference were 86% (76%, 96%), 99% (88%, 110%), and 90% (80%, 100%) for the 5, 20, and 40 mg capsules, respectively. The differences in the point estimates of the dose-normalized AUC(0-infinity), values for the 5, 20, and 40 mg capsules relative to the 10 mg phase-3 capsule were 1% or less, and the 90% confidence limits were all within 95-106%. These results indicate that stavudine exhibits linear pharmacokinetics and that the 5, 10, 20, and 40 mg capsules of stavudine are bioequivalent.