Bromate (BrO 3 -) in drinking water is formed when water containing bromide ion is treated with ozone, or it can be present in hypochlorite solutions produced from electrolysis of bromidecontaining salt (Weinberg et al, 2003;Moll & Krasner, 2002). Commercial bottled water produced with ozonation of desalted or groundwater sources could also contain elevated BrO 3 concentrations. Thus, BrO 3 is primarily introduced into drinking water during treatment.BrO 3 administered in drinking water produces cancer of the kidney and thyroid gland in male and female F344 rats and cancer of the testicular mesothelium in males Kurokawa et al, 1990). Among the species in which BrO 3 has been tested, rats are unusually sensitive to this compound. The lowest confidence limit for the BrO 3 dose that increases the incidence of tumors by 10% (LTD 10 ) in a male rat is estimated at 0.76 mg/kg/d, whereas the LTD 10 in mice is estimated at 3.3 mg/ kg/d and 33.1 mg/kg/d in hamsters (Gold et al, 2012). Although three target organs have been identified as sites of BrO 3 --induced tumors in rats, only kidney tumors have been observed in mice and hamsters. These large differences in potency and endpoints among species create considerable uncertainty for estimating cancer risk in humans on the basis of the animal data.Assuming BrO 3 to be a genotoxic carcinogen, the US Environmental Protection Agency (USEPA) estimated the cancer risk in humans to be 2 × 10 -4 (two in 10,000) additional cancers per lifetime at the maximum contaminant level (MCL) of 0.010 mg/L (USEPA, 2001). According to US drinking water standards, lifetime risk in the range of 10 -6 to 10 -4 (one in 1 million to one in 10,000) is generally considered safe and protective of health, but USEPA set the MCL goal (MCLG) at zero as a matter of policy on the basis of the assumption that BrO 3 is a genotoxic carcinogen. The state of California has re-evaluated this risk assessment by considering data from animals that were not included in the DeAngelo et al (1998) study but were subsequently made available for a revised risk analysis (OEHHA, 2009). The California public health goal derived from this analysis was 0.l µg/L. It can be anticipated that USEPA will consider revising the MCL in the six-year review that is under way.USEPA's Cancer Risk Assessment Guidelines (USEPA, 2005) provide criteria for establishing that a chemical acts by a nonlinear mode of action (MOA). These criteria are summarized in abbreviated form in the sidebar on page E710. Evidence supporting nonlinear extrapolation could include nonlinear relationships between external doses of a cancer-causing agent and doses that reach affected tissues or demonstration that the compound induces cancer by an MOA that is nonlinear at low doses (frequently referred to as having a threshold). Therefore, a determination that tumors induced by BrO 3 are the result of nongenotoxic effects could result in a nonzero MCLG.Risk assessments for bromate (BrO 3 -) in drinking water are based on linear extrapolation from the total incidenc...