In mice depleted of GSH by treatment with buthionine sulfoximine (BSO), methimazole (2-mercapto-1-methylimidazole, MMI) causes liver injury characterized by centrilobular necrosis of hepatocytes and an increase in serum alanine transaminase (SALT) activity. MMI requires metabolic activation by both P450 monooxygenase and flavin-containing monooxygenase (FMO) before it produces the hepatotoxicity. MMI and its analogues were examined for the ability to increase SALT activity in GSH-depleted mice. Saturation of the C-4,5 double bond in MMI resulted in a complete loss of hepatotoxicity. Similarly, ring fusion of a benzene nucleus to the C-4,5 double bond, forming 2-mercapto-1-methylbenzimidazole, abolished the toxic potency. As for MMI, 2-mercapto-1,4,5-trimethylimidazole, and 2-mercapto-1-methyl-4, 5-di-n-propylimidazole, the toxic potency decreased with the increasing bulk of the 4- and 5-alkyl substituents. Furthermore, methylation of the thiol group of MMI totally reduced its toxicity. These structural requirements and the known toxicity of thiono-sulfur compounds led us to the hypothesis that MMI would undergo epoxidation of the C-4,5 double bond by P450 enzymes and, after being hydrolyzed, the resulting epoxide would be then decomposed to form N-methylthiourea, a proximate toxicant. Before N-methylthiourea would produce toxicity, it would be further biotransformed to its S-oxidized metabolites mainly by FMO. Evidence for this hypothesis was provided by the facts that N-methylthiourea and glyoxal as the accompanying fragment were identified as urinary metabolites in mice treated with MMI and that N-methylthiourea caused a marked increase in SALT activity when administered to mice in combination with BSO.