In a dose-ranging study of amprenavir (formerly 141W94), an inhibitor of the protease enzyme of human immunodeficiency virus (HIV) type 1, single-dose and steady-state pharmacokinetic parameters were estimated from plasma samples collected on day 1 and during week 3, respectively. Amprenavir was administered on either a twice-daily (b.i.d.) or three-times-daily dosage schedule to 62 HIV-infected adults, 59 of whom had pharmacokinetic data. Log-log regression analysis (the power model) revealed that the steady-state area under the curve (AUC(ss)) and the maximum, minimum, and average concentrations at steady state (C(max,ss), C(min,ss), and C(avg,ss), respectively) increased in a dose-proportional manner over the 300- to 1,200-mg dose range. Steady-state clearance was dose independent. AUC(ss)/AUC(0-->infinity) decreased linearly with dose and correlated significantly with treatment-associated decreases in alpha(1)-acid glycoprotein. After 3 weeks, the dose of 1,200 mg b.i. d. provided a median amprenavir C(min,ss) (0.280 microg/ml) that was higher than the median in vitro 50% inhibitory concentration for clinical HIV isolates (0.023 microg/ml), even after adjustment for protein binding. The median amprenavir C(min,ss) was also greater than the estimated in vivo trough concentration calculated to yield 90% of the maximum antiviral effect (0.228 microg/ml) over 4 weeks. A pharmacodynamic analysis of the relationship between steady-state pharmacokinetic parameters and safety revealed headache and oral numbness to be the only side effects significantly associated with C(max). The pharmacodynamic relationship defined in this study supports the use of 1,200 mg b.i.d. as the approved dose of amprenavir.