Bruce Ames has an excellent Letter to the Editor in the October issue of Risk Analysis. Although he focuses on chemical carcinogens, he concludes that a linear cancer model is inappropriate and supports a possible threshold.