We aimed to assess the effect of coadministration and withdrawal of a potent cytochrome P450 3A (CYP3A) inhibitor (ritonavir) and a potent CYP3A inducer (St John's wort) on CYP3A enzyme activity in an open, fixed-sequence study design. We investigated the pharmacokinetics of midazolam: (i) at baseline, (ii) after a single dose of either St John's wort or ritonavir (each n = 6), (iii) after 14 days of coadministration of ritonavir (300 mg b.i.d.) and St John's wort (300 mg t.i.d.), and (iv) at 2 days after cessation of both St John's wort and ritonavir. Combined administration of inducer and inhibitor resulted in a predominance of enzyme inhibition: coadministration of St John's wort and ritonavir with intravenous administration of midazolam resulted in an increase in the area under the plasma concentration-time curve (AUC)(0-8 h) of midazolam to 180% of baseline value, whereas with orally administered midazolam, the AUC(0-6 h) increased to 412% of baseline value (P < 0.05 for each). After cessation of the coadministered drugs, the AUC(0-6 h) of orally administered midazolam decreased to 6% of the level observed during combined administration, and the AUC(0-8 h) of intravenously administered midazolam decreased to 33% of the values observed during combined administration (P < 0.001 for each). Induction may be unmasked after the withdrawal of a combination of a potent CYP3A inhibitor and a potent CYP3A inducer, leading to substantial drops in drug exposure of CYP3A substrates. This may require substantial dose adjustments, particularly of orally administered drugs.