Human immunodeficiency virus (HIV) protease inhibitors (PIs) are inhibitors of CYP3A enzymes, but the mechanism is poorly defined. In this study, time- and concentration-dependent decreases in activity as defined by maximum rate of inactivation (k(inact)) and inhibitor concentration that gives 50% maximal inactivation (K(I)) of CYP3A by amprenavir, indinavir, lopinavir, nelfinavir, ritonavir, and saquinavir were quantified using testosterone 6beta-hydroxylation as a marker for CYP3A activity with recombinant CYP3A4(+b(5)), recombinant CYP3A5, and pooled human liver microsomes (HLMs). All the PIs, except indinavir, displayed inactivation with CYP3A4(+b(5)) and HLMs. Ritonavir was the most potent (K(I) = 0.10 and 0.17 microM) and demonstrated high k(inact) values (0.32 and 0.40 min(-1)) with both CYP3A4(+b(5)) and HLMs. Ritonavir was not significantly depleted by high-affinity binding with CYP3A4(+b(5)) and confirmed that estimation of reversible inhibition was confounded with irreversible inhibition. For CYP3A5, nelfinavir exhibited the highest k(inact) (0.47 min(-1)), but ritonavir was the most potent (K(I) = 0.12 microM). Saquinavir and indinavir did not show time- and concentration-dependent decreases in activity with CYP3A5. Spectrophototmetrically determined metabolic intermediate complex formation was observed for all of the PIs with CYP3A4(+b(5)), except for lopinavir and saquinavir. The addition of nucleophilic and free aldehyde trapping agents and free iron and reactive oxygen species scavengers did not prevent inactivation of CYP3A4(+b(5)) by ritonavir, amprenavir, or nelfinavir, but glutathione decreased the inactivation by saquinavir (17%) and catalase decreased the inactivation by lopinavir (39%). In conclusion, all the PIs exhibited mechanism-based inactivation, and predictions of the extent and time course of drug interactions with PIs could be underestimated if based solely on reversible inhibition.