Heavy metals are major soil pollutants since a lot of former industrial soils are polluted by these contaminants. In the context of risk assessment of contaminated sites, they are of particular concerns because of their toxicity toward human beings. Nevertheless, the vadose zone of the soil is not taken into consideration in this kind of studies though this is where the pollution enters the soil. That is why mechanisms responsible for trace element release in the unsaturated zone of the soil have to be understood. In addition, tools and methods to measure and put forward interactions between contaminants and the solid matrix have to be provided. Thus the aim of this study was to characterize As transport in soils in the field and to put forward mechanisms responsible for its migration in the unsaturated zone of the soil by setting up an original laboratory experiment. The study site was an in-service wood preserving facility site, located in the north east of France. A regulatory monitoring of the groundwater quality was initiated by a prefectural order in 1998 and 4 water wells are monitored. Water samples are taken twice a year. Soils were contaminated with Cr, Cu and As and we were mainly interested in As release since it was the only trace element measured in groundwater at a concentration above the EU guidelines for drinking water quality. Soils were sandy clay loam. On site, the mean depth of the water table was 6 m from the ground surface; the water table level variation was about 1 m over a year. Soils were sampled and analyze for physico-chemical characteristics and total As content and were submitted to a sequential extraction. In addition, two columns (30 cm high and 12 cm in diameter) were filled with 2 sub-samples of the As contaminated soil (202 and 253 mg/kg) and As release was studied as it could occur on site. Two main phenomena were simulated during these experiments: rain water infiltration and an increase in the water table level. Field studies in combination with a statistical analysis showed that As release was highly correlated to the water table level. After the results of the laboratory experiments, it was shown that As concentrations at the outlet of the two columns were constant over time. Given that and the results of the sequential extraction carried out on this soil, As release could mainly occur from the soluble and exchangeable fractions of As. Arsenic transport modeling is in progress in order to confirm this hypothesis.