The installation of hundreds of thousands of shallow tube wells (STW) in Bangladesh over the last three decades supplied “safe” drinking water for the people and, also, has been an important contributor to the tripling of the annual production of rice, the staple food crop. However, within two decades, As contamination of the all important groundwater began to show up and presently, many millions of Bangladeshis face a two-way risk of exposure to As, directly through drinking water and indirectly through food crops grown on soils contaminated by high-As groundwater irrigation. In early 2002, we began a systematic study of the nature and extent of the As problem in Bangladesh agriculture and food. One of our objectives has been to assess the As contamination status of irrigation waters and soils throughout the country and its impact on food quality. Here, we report some findings of a national survey on As in agricultural systems conducted in 2004-2005. During this period, irrigation water, soil and crop samples (both rice and non-rice crops) from 184 unions of 92 of the 450 thanas (a thana is the smallest administrative unit in Bangladesh) across the country were collected and analyzed for As using the HG-AAS procedure. The national survey showed very wide ranges of As contamination in waters, soils and crops across Bangladesh. Seventy-seven percent of the 355 STWs (irrigation tubewells) sampled had a relatively low As content, <100 µg/L, 15% had a high As level (100-200 µg/L) and the remaining 8% had very high As, >200 µg/L. The As content in the soils ranged from negligible to 64 mg/kg, with an average of 6.5 mg/kg for 394 soil samples. About 55% of the samples had an As content of <5 mg/kg, but at least 25% of the soils had high As levels of 10 to >20 mg/kg. Such high As levels in soils may be undesirable for crop production, especially rice, because chemical and biochemical conditions in the traditional wetland rice production system favor the mobilization of As from the soil. Out of the 345 samples of irrigated rice (dry season rice, locally called Boro), the grain As content in about 50% was greater than 0.2 mg/kg, with a range of 0.04 to 1.10 mg/kg and an average value of 0.31 mg/kg. The corresponding wet season rain-fed rice had a mean grain As content of 0.17 mg/kg. Because rice is the staple food in Bangladesh and it is consumed in large quantities (450 g/adult/day), As contaminated rice could be an additional human health risk along with As-charged drinking water in many areas of Bangladesh. The As contents in the human edible parts of 70 different non-rice crops, like wheat, maize, leafy vegetables, tomato, chili, beans, etc. were also determined. While As in the upland cereals, maize and wheat, was found to be negligible, most other crops, especially, the leafy vegetables and tubers appeared to accumulate As in amounts double or triple that of rice. However, the risk from these crops would be low since the human consumption of the edible parts is much lower than that of rice. An important finding was the lack of straightforward interrelationships among the As contents of irrigation waters, soils and crops. Individual and interactive impacts of irrigation water As, “background” soil As, land type and land form, flooding patterns, chemical and mineralogical processes, cropping sequences, water management, etc. need to be studied in details for information contributing to local and regional delineation of the actual and potential As hazard. Future research should focus on water-soil-crop management systems to minimize the As risk in Bangladesh agriculture and food.