Decrease of As authorized level in drinking waters to 10 μg.l-1 requires developing treatments with low costs. In water, arsenic is mainly found as inorganic forms: As(V) and As(III), the latter being more toxic and mobile. Water treatments are therefore more effective to remove As(V) and thus, a pre-oxidation step is generally necessary. Biological treatment using As(III)-oxidising bacteria, followed by As adsorption on Fe0 coated sand bed reactor, could be a promising treatment process. Thiomonas arsenivorans, an autotrophic bacteria isolated from a gold mine in France, was used to oxidise As(III) in fixed bed bioreactors. The objectives of this work were to study the influence of filling material (sand or Fe0-coated sand) on bacteria development and to compare performances obtained for biological oxidation in a single oxidation-adsorption reactor or in an oxidation reactor followed by an adsorption one. Experiments were carried out in three thermostated reactors (12°C). C1 was filled with sand (sieved at 1-2 mm) and 1% Fe0, C2 was identical but inoculated with Thiomonas arsenivorans and C3 was filled with sand only and inoculated. Bed volume and hydrodynamic parameters of all reactors were characterised. After inoculation of C2 and C3, the three bioreactors were continuously up-flow fed with a synthetic water containing 10 mg.l-1 As(III). This high concentration was chosen in order to favour bacteria development. During experiments, water samplings were performed at several heights along the reactors for arsenic speciation analysis using HPLC-HG-AFS. At the end of the experiment, the contents of bioreactors were collected and divided into different parts according to their height in the reactor in order to study bacterial distribution. For each collected fraction, batch experiments were performed to study kinetics of As(III) bacterial oxidation. Bacteria diversity was also analysed on these fractions by PCR-DGGE.