A sensitive and robust method for the determination;of eight inorganic and organic arsenic species by ion-exclusion liquid chromatography combined with inductively coupled plasma mass spectrometry (LC-ICP-MS) and. LC-hydride generation-ICP-MS (LC-HG-ICP-MS) is described. The species are arsenite (As-III), arsenate (As'), monomethylarsonic acid (MMAs), dimethylarsinic acid (DMAs), arsenobetaine (AsB), trimethylarsine oxide (TMAsO), tetramethylarsonium salt (TMAs) and arsenocholine salt (AsC). A good separation of seven arsenic species, As-III, As-v, MMAs, DMAs, AsB, TMAsO, and AsC or TMAs, was achieved by using an ion-exclusion column packed with a carboxylated methacrylate resin and 0.35 mmol l(-1) of a sodium sulfate solution adjusted to pH 3.8 as the mobile phase. The detection limits of the eight arsenic species obtained by LC-ICP-MS ranged from 0.067 to 0.34 ng As ml(-1) using an injection volume of 50 mul. A hydride generation technique improved the detection limits of As-III, As-v, MMAs, DMAs and TMAsO to 0.016-0.075 ng As ml(-1) while AsB, TMAs and AsC were not detectable. The relative standard deviations of five replicates of a standard of each arsenic species by the former method ranged from 1.3 to 3.3% and those by the latter method, from 1.8 to 3.1%. The proposed methods were successfully applied to the determination of five arsenic species in human urine and six species in an extract from tuna fish tissue. The only necessary pretreatment for these analyses involved filtration with a 0.45 mum membrane and the ArCl polyatomic interference, due to a large amount of chloride in the biological samples on As-75 measurement, was eliminated by the described LC separation. In addition, serious deterioration in column performance and a decrease in the sensitivity of ICP-MS were not observed during the experimental period of five months. The LC-ICP-MS and LC-HG-ICP-MS methods were validated by analyzing reference samples of human urine and tuna fish tissue.