High levels of As in drinking-water have led to an unresolved health crisis in Bangladesh. Groundwater rich in As is also increasingly used for irrigation of paddy rice fields during the dry season. The As transferred to the rice fields with irrigation water may accumulate in soils, decrease soil fertility, and enter the food chain via plant uptake. This may lead to additional long-term risks for environmental and human health. Our objective is to investigate the biogeochemical cycling of As in the paddy soil - rice system and to evaluate the potential risk of As accumulation in soils. In order to quantify As fluxes, spatial and temporal variations in As concentrations have to be considered. Understanding As transformations by biogeochemical processes requires information about the As speciation at the soil profile to micrometer scale. We therefore designed a field study to investigate the As cycle in a selected study area located 30 km south of Dhaka. The field site comprises 18 paddy fields irrigated by one well, delivering water with 400 µg As/L. Soil samples are collected twice a year, before (December/January) and after (May) the dry season in which irrigated rice is grown. Soil samples are collected on a regular grid at several depths (0-45 cm). Plant samples are taken shortly before harvest (April). In January 2005, 12 fields were sampled on a 20x20 m grid (2-9 samples per field). One field was sampled in greater detail using a more detailed grid sampling scheme (38 samples). Additional samples were taken from the irrigation channels. In May 2005, soil sampling was restricted to two fields (0.28 ha; 0.33 ha) situated at different distances from the well and to the channels leading into these fields. Soil profiles were sampled with a high vertical resolution at both dates. All samples were oven-dried at 60ºC for soil analyses. Our results obtained so far demonstrate that the input of As into the paddy fields does not depend on their distance from the irrigation well, but that the As concentration in each individual field strongly decreases with increasing distance to the water inlet. Arsenic concentrations measured at the end of the irrigation season show that the As input was about five times higher near the water inlet than at the opposite side of the field. This heterogeneous input of As is most likely related to the oxidation of Fe(II) to Fe(III) and of As(III) to As(V) in the initially anoxic irrigation water. In contact with air, Fe(II) is oxidized to Fe(III), resulting in formation of Fe(III)-hydroxide colloids. As(V) strongly adsorbs to these iron hydroxide particles. Subsequent colloid aggregation and removal by sedimentation lead to As deposition onto the soil surface. This is supported by the vertical distribution of total As in the soil profiles, which shows a strong As enrichment in the top few centimeters at the end of the irrigation season, and relatively low As concentrations before the onset of irrigation. Phosphorus shows a similar distribution pattern, indicating analogous removal mechanisms. Since the As concentrations in May were generally higher compared to January and the input was mostly observed in the upper soil layers, long-term As accumulation appears to be counteracted by biogeochemical pathways leading to As removal. As may be leached or transported into deeper soil layers and laterally with flood water during the monsoon season. In addition, As concentrations in the topsoil are homogenized annually by soil puddling.