The aim of the present study was to estimate the time course change in cytochrome P450 3A (CYP3A) activity during repeated doses of erythromycin. Twelve healthy male volunteers participated in this randomized, 4 x 4 Latin square design study. The pharmacokinetics of a single oral dose of midazolam, a probe for CYP3A activity, were assessed in 4 conditions: (1) midazolam (5 mg) without erythromycin (EM0), (2) erythromycin 2 days + midazolam (2.5 mg) (EM2), (3) erythromycin 4 days + midazolam (2.5 mg) (EM4), and (4) erythromycin 7 days + midazolam (2.5 mg) (EM7). The dose of erythromycin was 800 mg/d. Erythromycin produced a 2.3-, 3.4-, and 3.4-fold increase in dose-corrected area under the curve of midazolam for EM2, EM4, and EM7, respectively, as compared with EM0 (P <.05/6). A significant prolongation of terminal half-life was observed in EM4 and EM7. The relationship between the duration of erythromycin treatment and total clearance of midazolam indicated that a plateau level of CYP3A inhibition can be achieved by 4 days or more of erythromycin treatment. The repeated treatment with erythromycin yields CYP3A inhibition in a duration-dependent manner. A 4-day course of erythromycin treatment produces 90% or more of the maximal inhibition of CYP3A in humans.