Regulatory focus on quantifying risk of disease or death from exposure to hazardous substances via monotonic dose-response models has downplayed or even rejected potential benefits to human health from exposures to low (sub-threshold) doses, and thus represented by either U-shaped or J-shaped models. On the other hand, most environmental health policy hypothesizes, without firm evidence, that cancer risk is proportional to exposure at low doses of current ambient exposures. An acceptable exposure is determined by either setting a somewhat arbitrary ;acceptable' level of risk, such as one in a million excess individual lifetime cancer risk or, in the case of several types of animal toxicological test results, applying multiplicative safety factors to a specific concentration, generally derived from a benchmark dose or NOAEL. This seemingly precautionary approach is questionable in light of much experimental evidence indicating protective effects of exposure at low doses - U-shaped or J-shaped models. We demonstrate that incorporating the possibility of hormesis into regulatory decision-making is precautionary, while use of default results in policy conflicts with precaution.