A large number of drinking water supplies worldwide have greater than 50 microg l(- 1) inorganic arsenic in drinking water, and there is increasing pressure to reduce concentrations. Few studies have specifically considered low concentrations of arsenic in water supplies and the significance of other factors which may contribute to increased exposure. This study aimed to investigate risk factors for increased urinary inorganic arsenic concentrations, in a population exposed to 10 - 100 microg l(- 1) of arsenic in drinking water, as well as a control population with lower arsenic concentrations in their drinking water. Inorganic arsenic in urine was used as the measure of exposure. The median drinking water arsenic concentration in the exposed population was 43.8 microg l(- 1) (16.0 - 73 microg l(- 1)) and less than the analytical limit of detection of 1 microg l(- 1) (<DL) (range<DL-8.0 microg l(- 1)) in the control group. The geometric mean urinary inorganic arsenic concentration for the exposed group was 4.24 microg l(- 1) (range<DL-18.8 microg l(- 1)) and for the control group was 1.18 microg l(- 1) (<DL-4.49 microg l(- 1)). In a random effects linear regression model, drinking water was the significant predictor of urinary inorganic arsenic concentrations with factors such as age, season and drinking water consumption important risk factors. These results show that concentrations of arsenic in drinking water, even at lower concentrations, make an important contribution to exposure. Further work is required to define the potential for absorption at these lower levels.