In rats exposed to arsenite (AsIII) or arsenate (AsV), the biliary excretion of arsenic depends completely on availability of hepatic glutathione, suggesting that both AsIII and AsV are transported into bile in thiol-reactive trivalent forms (Gyurasics et al. [1991], Biochem. Pharmacol. 42, 465-468). To test this hypothesis, the bile and urine of bile duct-cannulated rats injected with AsIII or AsV (50 micromol/kg, iv) were collected periodically for 2 h and analyzed for arsenic metabolites by HPLC-hydride generation-atomic fluorescence spectrometry. Arsenic was excreted predominantly into bile in AsIII-injected rats, but the urine was the main route of excretion in AsV-exposed rats. Injected AsIII was excreted in urine practically unchanged, whereas both AsV and AsIII appeared in urine after administration of AsV. Irrespective of the arsenical administered, the bile contained 2 main arsenic species, namely AsIII and a hitherto unidentified metabolite. Formation of this metabolite could be prevented by pretreatment of the rats with the methylation inhibitor periodate-oxidized adenosine, indicating that it is a methylated arsenic compound. This metabolite could be converted in vitro into monomethylarsonic acid (MMAsV) by oxidation, whereas synthetic MMAsV could be converted into the unknown metabolite by reduction. Consequently, this biliary metabolite of both AsIII and AsV is monomethylarsonous acid (MMAsIII), a long-hypothesized, but never identified, intermediate in the biotransformation of AsIII and AsV. Although MMAsIII is thought to be formed from an oxidized precursor, rats injected with MMAsV did not excrete MMAsIII. In summary, the inorganic arsenicals investigated are transported into bile exclusively in trivalent forms, namely as AsIII and MMAsIII, but are excreted in urine in both tri- and pentavalent forms. Identification of MMAsIII is signified by the fact that this metabolite is more toxic than AsIII and AsV and thus formation of MMAsIII represents toxification of inorganic arsenic.