OBJECTIVE: To measure the in vivo variations of CYP3A activity induced by anti-HIV drugs in human immunodeficiency virus (HIV)1-positive patients. METHODS: A low oral dose of midazolam (MID) (0.075 mg) was given to the patients and the 30-min total 1-OH midazolam (1-OHMID)/MID ratio was determined. Patients were phenotyped either before the introduction of antiretroviral treatments (control group, 90 patients) or after a variable period of antiretroviral treatment (56 patients). Twenty-one subjects underwent multiple phenotyping tests (before and during the course of the treatment). RESULTS: The median MID ratio was 3.51 in the control group (range 0.20-14.6). It was 5-fold higher in the group with efavirenz (28 patients; median, range: 16.0, 3.81-367; P < 0.0001), 13-fold lower with nelfinavir (18 patients; 0.27, 0.06-36.3; P < 0.0001), 17-fold lower with efavirenz + ritonavir (three patients; 0.21, 0.05-0.47; P = 0.006), 50-fold lower with ritonavir (four patients; 0.07, 0.06-0.17; P = 0.0007), and 7-fold lower with nevirapine + (ritonavir or nelfinavir or grapefruit juice) (three patients; 0.48, 0.03-1.83; P = 0.03). CYP3A activity was lower in the efavirenz + ritonavir group (P = 0.01) and in the ritonavir group (P = 0.04) than in the nelfinavir group, although already strongly inhibited in the latter. CONCLUSION: The low-dose MID phenotyping test was successfully used to measure the in vivo variations of CYP3A activity induced by antiretroviral drugs. Efavirenz strongly induces CYP3A activity, while ritonavir almost completely inhibits it. Nelfinavir strongly decreases CYP3A activity, but to a lesser extent than ritonavir. The inhibition of CYP3A by ritonavir or nelfinavir offsets the inductive effects of efavirenz or nevirapine administered concomitantly. Finally, no induction of CYP3A activity was noticeable after long-term administration of ritonavir at low dosages (200 mg/day b.i.d.) or of nelfinavir at standard dosages (2,500 mg/day b.i.d.).