In sulfidic environments, hydride generation-based approaches are not suitable for arsenic determination because thioarsenates which can constitute the predominant arsenic species under these conditions (>80% of total arsenic) are completely ignored. Sample acidification for preservation or during hydride generation leads to loss of total inorganic arsenic due to precipitation of arsenic-sulfur phases. Total concentrations can be determined correctly using 1% potassium iodide as prereducing agent while with L-cysteine (0.16 mol L-1), transformation of tetra-, tri-, and dithioarsenate to arsenite remains incomplete. By decreasing the original sample pH, hydride generation destroys thioarsenate species distribution because only monothioarsenate is stable over the whole pH range. Dithioarsenate transforms to arsenite below pH 4. Tetrathioarsenate transforms to trithioarsenate( pH 11.9) which subsequently transforms to arsenite (pH 5.6), followed by precipitation of arsenic-sulfur phases below pH 5. It is thus impossible to determine thioarsenates by hydride generation. The “As(III)”-fraction contains tetra-, tri-, and some dithioarsenate as well as arsenite, while monothioarsenate is determined with arsenate as “As(V)”. Different analytical setups have substantial impact on thioarsenate hydride-generation behavior, thus provide little comparability and render reinterpretation of existing arsenic hydride-generation speciation data from sulfidic environments impossible. In natural geothermal water samples from Yellowstone National Park, total arsenic concentrations determined by ICP-MS and by HG-AFS with prereductant agreed well (<6% relative difference). Speciation results deviated from the behavior predicted for thioarsenates from laboratory experiments, probably due to matrix effects.