Lamotrigine, carbamazepine (CBZ), and phenytoin (PHN) are associated with a relatively high incidence of idiosyncratic drug reactions (IDRs). Evidence suggests that such reactions are caused by reactive metabolites and are immune-mediated. One hypothesis that can link reactive metabolites with an immune-mediated reaction is the danger hypothesis, i.e. cell stress is important for the induction of an immune reaction. CBZ causes neutropenia suggesting that the reactive metabolite is formed by myeloperoxidase (MPO), which is the major oxidative enzyme in neutrophils. A major hepatic metabolite of CBZ is the 3-OH-CBZ and that of PHN is 4-OH-PHN. We found that 3-OH-CBZ and 4-OH-PHN are further oxidized by MPO/H2O2 to dimers, strong evidence for a free radical intermediate. Bioactivation of 3-OH-CBZ and 4-OH-PHN in the presence of model proteins lead to covalent binding, and MS/MS fragmentation revealed that the covalent binding site is tyrosine.Although lamotrigine can cause agranulocytosis, the major IDR is skin rash. Lamotrigine is oxidized by MPO to a reactive N-chloro metabolite, which can be trapped with N-acetylhistidine; however, the amount of covalent binding in the presence of MPO was quite low. The amount of covalent binding to hepatic microsomes was even less. Despite testing other potential pathways mediated by enzymes present in the skin, such as sulfation of lamotrigine-N-oxide, no pathway was found that could reasonably be linked to reactive metabolite formation in the skin. It is possible that lamotrigine-induced skin rash is caused by the parent drug.Free radicals can generate ROS and other forms of cell stress consistent with the danger hypothesis, and we found that 3-OH-CBZ, but not 4-OH-PHN, generated reactive oxygen species (ROS) in a peroxidase-containing cell line. We found that CBZ and PHN treatment induced genes of protective enzymes related to the Keap-Nrfl-ARE signal pathway, such as glutathione transferase, and stress proteins, such as heat shock proteins. It also induced IL-6, a major mediator of the acute phase response, and increased urinary isoprostane levels, a marker of lipid peroxidation. We postulate that bioactivation of 3-OH-CBZ and 4-OH-PHN to free radicals by peroxidases may play an important role in CBZ- and PHN-induced IDRs.