(FORTUNE Magazine) – Toxic-tort lawyers aren't going to like this: Evidence is growing that most hazardous chemicals, as well as radiation, not only are harmless at low doses--but may actually do a body good. Scientists who study this mind-bending effect, called hormesis, have quietly marshaled a wealth of examples showing that it's real. And it appears to be more the rule than the exception--long-ignored signs of the phenomenon have been unearthed in hundreds of studies on everything from toxic heavy metals to dread carcinogens like dioxin.

Hormesis is pure poison to the conventional wisdom in toxicology. It contradicts the idea that carcinogenic chemicals pose risk at any dose, no matter how low. That concept has dominated regulatory thinking since the 1970s, helping motivate billions of dollars of cleanups and countless lawsuits involving low-dose exposures. And if hormesis experts are right, the time-honored method for quantifying toxic risks at low doses--the levels typically encountered outside the lab--has all the accuracy of a fun-house mirror.

Here's the standard technique: Lab animals are exposed to megadoses of a toxin, causing pronounced effects that are readily measured. That yields guesstimates of the human effects at megalevels. To estimate the risk at low doses, regulators assume that the toxic effects fall in a straight line with the dose. Tumor incidence in rats vs. doses of saccharin, for instance, would be graphed on X and Y axes as a straight line. The handy line shows the purported risk at doses down to zero.

But scientists who go to the trouble of measuring actual toxic effects at low levels often observe a J-shaped "dose response" curve instead of a straight line. That means the risks many toxins pose at real-world levels have probably been exaggerated. The J-curve also suggests an idea that, at first blush, seems daft: Policies that foster small exposures to toxins might be better for public health than ones aimed at eliminating them.

Don't mess with this radical notion at home. On second thought, it may well be too late: Consider the toxins you've probably already taken at low doses in order to get their hormetic effects: a kidney poison known as vitamin D, a neurotoxin called caffeine, and that edifying solvent and fuel additive, ethyl alcohol--it seems that people who regularly imbibe small doses of alcohol, whether in red wine or other drinks, live longer.

Exercise fits the J-curve too. Moderate workouts are plainly beneficial--they can boost the immune system and lower the risk of heart disease. But overdoing it can suppress immune function and deplete internal stores of antioxidants, potentially leading to tissue damage from "free radicals." (Free radicals are highly reactive molecules generated in cells as they burn fuel--antioxidants neutralize their destructive power.)

To hormesis researchers, the rising interest in toxins' yin-yang effects seems like the end of 70 years of mass amnesia. Before 1930 the idea that low doses of poisons could be invigorating was mainstream science. Researchers reported that lightly dousing bacteria with germicides stimulated their growth, that a little arsenic revved up yeast metabolism, and that patients with bacterial infections benefited from low doses of X-rays.

Then quacks gave hormesis a bad name. It became associated with homeopathy, the bogus idea that diseases can be treated with tiny traces of toxins. (Homeopaths endorse much smaller doses than those that typically trigger hormesis.) Worse, radium-laced elixirs hit the market. Citing studies on the positive effects of low-dose radiation, makers of potions touted them as energizing cure-alls. In 1932 came a high-profile tragedy: Steel millionaire and playboy Eben Byers, a heavy user of an elixir called Radithor, suffered massive bone deterioration and died at 51 from radiation poisoning. Years ago the Wall Street Journal recounted the grisly case under a particularly memorable headline: THE RADIUM WATER WORKED FINE UNTIL HIS JAW CAME OFF.

Soon after, hormesis went the way of patent medicines. But the effect continued to turn up in toxicology research, generating little waves of interest. In 1990, University of Massachusetts public-health professor Ed Calabrese launched a hormesis revival by forming an international body to foster research on the phenomenon. Calabrese had stumbled onto hormesis as an undergraduate in the 1960s while studying the effects of a growth-inhibiting chemical on peppermint plants--they unexpectedly grew faster on low doses. He regarded the finding as a minor anomaly until the mid-1980s, when he heard that low doses of radiation had a similar effect. Suspecting a broad phenomenon was at work, he zeroed in on it. By the late 1990s he and colleague Linda A. Baldwin had ferreted out an impressive set of hormetic effects from the scientific literature--everything from the stimulation of tumor cells by low doses of a cancer drug called suramin to the speeding of insect development by low levels of cyanide.

Their work began ruffling feathers. A long-standing objection is that the definition of hormesis is hazy. Many toxicologists argue that the term should apply only to cases in which a known mechanism can explain the J-curve. For instance, a toxin might stimulate a cellular defense system that "overshoots" at low doses, affording lingering resistance to injury. Higher doses overwhelm the system, producing the J-shaped response.

This strict definition offers a major plus: It grounds hormesis in familiar science, making its radical implications plausible. Scientists have long known, for example, that zapping insects with mild doses of radiation revs up their cells' inner fix-it systems, which repair DNA and other molecules damaged by the rays. This hormetic response turns them into superbugs, hardened to larger doses that would otherwise kill. They live longer than unradiated insects too.

But Calabrese, the hormesis revolution's George Washington, prefers a more inclusive definition: Hormesis, he says, is defined by low-level stimulation and high-level inhibition--irrespective of underlying mechanisms. Thus, alcohol is hormetic in his view because it boosts longevity at low doses and lessens it at high ones, even though the mechanism by which it does so isn't understood.

Given all the hot buttons hormesis hits, you'd think it would have spawned endless vitriolic disputes. Instead it has mainly generated thoughtful debates. Credit for the civility goes mainly to Calabrese, who sees polarizing clashes as a waste. A wiry, soft-spoken bicycling enthusiast with tousled gray locks, Calabrese, 57, is the most disarming of contrarians. At a hormesis meeting he organized in 2000, a prominent colleague stood up and delivered a biting critique: Despite assembling many examples of hormesis, he complained, Calabrese had failed to establish whether a significant proportion of toxins manifest the effect. "I had to admit in front of everyone that he was right," says Calabrese. "After I licked my wounds, I called him up" and enlisted his help in designing a study to address the issue.

For the study, which turned out to be a landmark work, Calabrese and Baldwin spent a year poring over some 21,000 papers on pesticides, metals, industrial wastes, and other toxins. Few of the studies included enough data at low doses to show whether hormesis had occurred. Of the ones that did, though, hormetic responses were seen 2.5 times more often than responses going the other way.

That compelling result, detailed this year in Toxicological Sciences, put hormesis firmly on the map. Now Calabrese argues that regulators should use the J-curve, instead of straight lines, to estimate low-dose effects of toxins. That change would have momentous policy implications--and probably won't happen soon. But Calabrese sees many ways hormesis might be applied before it gains greater acceptance. For instance, it might well make sense to give low, protective doses of X-rays to people facing subsequent exposure to high doses of radiation, such as rescue teams coping with the aftermath of a terrorist "dirty bomb." The medical world probably isn't ready to entertain this far-out concept. But Tom Ridge might be.