BackgroundManganese is an essential nutrient, but in excess, can be a potent neurotoxicant. We previously reported findings from two cross-sectional studies on children, showing that higher concentrations of manganese in drinking water were associated with deficits in IQ scores. Despite the common occurrence of this neurotoxic metal, its concentration in drinking water is rarely regulated. ObjectiveWe aimed to apply a benchmark concentration analysis to estimate water manganese levels associated with pre-defined levels of cognitive impairment in children, i.e. drop of 1%, 2% and 5% in Performance IQ scores. MethodsData from two studies conducted in Canada were pooled resulting in a sample of 630 children (ages 5.9–13.7 years) with data on tap water manganese concentration and cognition, as well as confounders. We used the Bayesian Benchmark Dose Analysis System to compute weight-averaged median estimates for the benchmark concentration (BMC) of manganese in water and the lower bound of the credible interval (BMCL), based on seven different exposure-response models. ResultsThe BMC for manganese in drinking water associated with a decrease of 1% Performance IQ score was 133 μg/L (BMCL, 78 μg/L); for a decrease of 2%, this concentration was 266 μg/L (BMCL, 156 μg/L) and for a decrease of 5% it was 676 μg/L (BMCL, 406 μg/L). In sex-stratified analyses, the manganese concentrations associated with a decrease of 1%, 2% and 5% Performance IQ in boys were 185, 375 and 935 μg/L (BMCLs, 75, 153 and 386 μg/L) and 78, 95, 192 μg/L (BMCLs, 9, 21 and 74 μg/L) for girls. ConclusionStudies suggest that a maximum acceptable concentration for manganese in drinking water should be set to protect children, the most vulnerable population, from manganese neurotoxicity. The present risk analysis can guide decision-makers responsible for developing these standards.