In this study, the authors developed a sensitive and reproducible method for urinary beryllium and to use this method to establish levels in workers at an aluminium smelter and in unexposed persons. A method was developed for urinary beryllium using a Thermo ICPMS Series 1, which was used to determine beryllium concentrations in urine from 62 people with no known occupational exposure to beryllium and 167 workers with potential exposure to beryllium at an aluminium smelter, where beryllium exists as an impurity in the bauxite ore. The analytical method has a detection limit (based on three times the background equivalency concentration of the blank) for beryllium in urine of 6 ng/L. The mean and 90th percentiles of urinary beryllium for workers were 19.5 and 42.0 ng/L and compared with 11.6 and 20.0 ng/L in people not occupationally exposed to beryllium. Statistical analysis using mixed effects models showed that workers had 47% (in 135 paired samples) higher levels of urinary beryllium at the end of the working week compared to the start of week and that the workers who smoked also had significantly higher levels of urinary beryllium compared to those that did not smoke. In addition, there was a statistically significant difference between workers and controls in urinary beryllium concentrations not correlated for creatinine. The authors concluded that a sensitive and reliable analytical method was developed for urinary beryllium by ICPMS. The workers in this study were exposed to beryllium at very low levels.