There were extensive public hearings and research into the epidemiology of arsenic in drinking water, including cost-benefit analyses of how much lowering the standard would cost vs. how many cancer cases would be prevented.
If you're REALLY interested, two of the rule-making reports are linked to the Web page below. They are very long. But they show that the Clinton administration didn't pick 10 ppb "out of their booties". They also estimate that for the 10 ppb (or micrograms/liter) standard, the costs of implementation are about the same as the savings due to the reduced number of cancer cases and other health improvements. Look at Table III.E-7 in the "Final Rule" report.
That is absolute BS.
If you want the truth, follow this link. Here is an excerpt which clearly shows that the NRC (which has NO National Academy of Science members) deliberately ignored science and forced adoption of the absurd and arbitrary 10 ppb limit.
Some scientists advocate use of mathematical models to predict cancer risk from these lower levels of arsenic exposure. The assumptions used in these models are critical.
The EPA's model assumes that any exposure to arsenic increases cancer risk and that cancer risk increases in a linear fashion with increasing exposure. Other models assume there is a "safe" level of exposure or are "sublinear" with increases in cancer risk that are negligible at low doses. Virtually all known toxicological processes follow a sublinear model.
Based on a variety of biochemical, toxicological and human study findings, the subcommittee stated in its report that only the sublinear models were plausible. But because subcommittee could not agree on which sublinear model was correct, the consensus forced by the NRC/EPA process — incredibly — was to opt for the EPA's linear model, the very model the arsenic subcommittee decided was wrong.
The linear model forced by the NRC process forces the adoption of the low standard because the effects at lower levels are artificially inflated. The correct model (the sublinear model) gives a realistic standard because the low level noneffect of arsenic is properly represented.