Taking advantage of mice deficient in gamma-glutamyl transpeptidase that are unable to metabolize glutathione (GSH), we have identified two previously unrecognized urinary metabolites of arsenite: arsenic triglutathione and methylarsenic diglutathione. Following administration of sodium arsenite to these mice, approximately 60-70% of urinary arsenic is present as one of these GSH conjugates. We did not detect the dimethyl derivative, dimethyl arsenic GSH; however, dimethyl arsenic (DMAV) represented approximately 30% of urinary arsenic. Administration of buthionine sulfoximine, an inhibitor of GSH synthesis, to wild-type mice reduced urinary arsenic excretion by more than 50%, indicating the GSH dependence of arsenic metabolism, transport, or both. Rodents deficient in three known ABC family transporters (MRP1, MRP2, and MDR1a/1b) exhibited urinary arsenic levels similar or greater than those in wild-type rodents; however, administration of MK571, an MRP inhibitor, reduced urinary arsenic excretion by almost 50%. MK571-treated mice showed approximately 50% reduction of AsIII, MMAV, and AsV as compared to untreated wild-type controls, while DMAV levels were unchanged. These findings suggest that arsenic excretion is in part dependent on GSH and on an MRP transporter other than MRP1 or 2.