A field survey was conducted in arsenic impacted and non-impacted paddies of Bangladesh to assess how arsenic levels in rice (Oryza sativa L.) grain are related to soil and shoot concentrations. Ten field sites from an arsenic contaminated tubewell irrigation region (Faridpur) were compared to 10 field sites from a non-affected region (Gazipur). Analysis of the overall data set found that both grain and shoot total arsenic concentrations were highly correlated (P<0.001) with soil arsenic. Median arsenic concentrations varied by 14, 10 and 3 fold for soil, shoot and grain respectively comparing the two regions. The reason for the sharp decline in the magnitude of difference between Gazipur and Faridpur for grain arsenic was due to an exponential decline in the grain/shoot arsenic concentration ratio with increasing shoot arsenic concentration. When the Bangladesh data were compared to EU and US soil-shoot-grain transfers, the same generic pattern could be found with the exception that arsenic was more efficiently transferred to grain from soil/shoot in the Bangladesh grown plants. This may reflect climatic or cultivar differences.