The stability of four major arsenic (As) species during application of the BARGE (Bioaccessibility Research Group of Europe) unified bioaccessibility method (UBM) has been assessed. The concentrations of As species in the UBM gastric and gastro-intestinal (gastric + intestinal) phases were determined using HPLC-ICP-MS whilst the total As content in the samples was determined using ICP-MS alone. The arsenic species studied were arsenite As(III), arsenate As(V), dimethylarsinic acid (DMA) and monomethylarsonic acid (MMA). These species were separated in 10 min using an anion exchange column (Hamilton PRP-X100) with a mobile phase containing 20 mmol L-1 NH4H2PO4/1% methanol (pH 6.0). The recoveries of arsenic species spiked into the gastric and gastro-intestinal fluids were in the range 90-108%. No interconversion between As species was observed as a result of applying the BARGE UBM, which is a particularly important finding for the reliability of As(III) measurements. The accuracy of the BARGE UBM for in vitro extractable As(V) was verified using British Geological Survey (BGS) guidance material 102 (an ironstone soil). For a commercial rice sample, the bioaccessibility sequence of As was DMA > As(III) > As(V) for the gastric phase and As(III) > DMA > As(V) for the gastro-intestinal phase.