In vitro, erlotinib (0-30 µmol/l) and C-labelled midazolam (MDZ) (5 µmol/l) were incubated with human liver microsomes; separately, microsomes were preincubated with erlotinib (10 µmol/l) before the addition of MDZ. Results showed a time-dependent inhibition of MDZ metabolism by erlotinib, with a Ki of 7.5 µmol/l and an inactivation rate constant of 0.009/min. Patients with cancer (n=24) received a single oral dose of 7.5 mg MDZ and a single intravenous dose of 3 µCi [C-N-methyl] erythromycin on days 1, 8, 14 and 21. Patients also received 150 mg oral erlotinib daily from day 8 to day 14. Plasma concentrations of erlotinib and OSI-420 were determined on days 8 and 14; MDZ and 1'-hydroxymidazolam were determined on days 1, 8, 14 and 21. Coadministration of erlotinib resulted in a 4 and a 16% increase in CO2 on days 8 and 14, respectively, after the administration of erythromycin. The mean AUC0-last of MDZ decreased 17 and 34% after erlotinib treatment on day 8 and day 14, respectively. The half-life of MDZ and the AUC ratio of 1'-hydroxymidazolam to MDZ were not significantly changed. Although erlotinib may be a weak mechanism-based irreversible inhibitor of CYP3A4 in vitro, in vivo, erlotinib did not inhibit CYP3A-mediated metabolism, as determined by the erythromycin breath test and the MDZ pharmacokinetics. The mechanism for reduced exposure of MDZ is unclear, but may be because of an increase in intestinal metabolism or decreased absorption. These findings suggest that coadministration of erlotinib may not result in clinically relevant increases in exposure of CYP3A substrates.