Page 9

EPA MCL for arsenic in drinking water of 50 µg/L does not achieve EPA's goal for public-health protection and, therefore, requires downward revision as promptly as possible.

To improve future risk characterizations, the following are recommended: Sensitivity analyses should be conducted to determine whether the results, including the way exposure concentrations are grouped together, are sensitive to the choice of model. The potential effect of measurement error and confounding on the dose-response curve and associated confidence limits should be further addressed.

To assist in the application of cancer data observed in different populations to cancer risks predicted for the United States, information on nutritional factors in study populations that pertains to susceptibility to arsenic-induced cancer should be investigated.

Modeling of epidemiological data should not be limited to the multistage Weibull model. Other models, including those which incorporate information from an appropriate control population, should be considered. The final risk value should be supported by a range of analyses over a broad range of feasible