It wasn't science, but Raymond Hayducka did his risk-benefit calculations over a beer in a smoke-filled Newark, N.J., taproom, while a dispirited-looking stripper cracked her gum in time to the bump and grind.

Outside, moon-suited technicians from the Environmental Protection Agency were carefully scraping up soil samples at the site of an abandoned chemical plant, looking for traces of dioxin, a chemical so toxic that health officials consider it unwise for humans to be exposed to any amount.

But Hayducka, taking an exaggerated breath and thumping his chest, said confidently, "I worked in the chemical plant for 16 years and I've never been sick. I won't waste my time worrying about it."

It was a swift and simple mental calculation: the certainty of a weekly paycheck weighed against an unproven physical threat.

As an analytical technique, however, it was not much cruder than the ones government officials use daily in weighing the highly visible benefits of the chemical revolution against its pervasive -- and little understood -- risks.

Risk-benefit analysis has been around for years, the tool of stockbrokers, insurance companies and speculators. But more recently, it has become an increasingly important handmaiden of government regulators, applied to sensitive questions involving health and safety.

The new emphasis has opened a Pandora's box of questions, many of them revolving around the imponderables of morality, the value of human life, and the right of free men and women to accept risk versus the government's obligation to protect citizens from risks thrust upon them involuntarily.

The chemical risk, at its most extreme, was starkly evident in Bhopal, India, where a poison-gas leak from a pesticide plant killed more than 2,000 people and injured 100,000 more in a matter of hours.

Investigators have not determined whether the catastrophe in Bhopal was preventable or whether it was the kind of industrial accident that will happen under the most restrictive safety practices.

But the tragedy served to sharpen debate on a broader question: How well does society understand the risks of chemicals that do not come as a suffocating cloud in the dark of night but as an invisible and inescapable part of our daily lives?

The debate is most intense inside the EPA, which has taken the lead in promoting risk-benefit analysis. EPA Administrator William D. Ruckelshaus addressed the issue squarely in one of his first speeches after returning to the agency in 1983.

"We must assume that life now takes place in a mine field of risks from hundreds, perhaps thousands, of substances," Ruckelshaus told the National Academy of Sciences. "No more can we tell the public: 'You are home free with an adequate margin of safety.' "

Against the staggering size of the regulatory task, the argument has undeniable intellectual force. More than 60,000 chemicals are in commercial use in the United States, and the"industry produces 1,000 new ones every year. There are 35,000 or more pesticides, 8,600 food additives, 3,400 cosmetic ingredients.

Before World War II, U.S. industry produced less than 10 billion pounds of synthetic organic chemicals each year. By 1980, the figure was 350 billion pounds and still climbing.

To grapple with a problem of that magnitude, Ruckelshaus argues, the government needs a uniform way of measuring risks and a statutory formula for balancing them against the economic value of a substance and the costs of controlling it.

"If the government gets better at both assessing and managing risk, and we do it accurately, people will have a better sense of the risks and will act sensibly," he said.

But even within the EPA, opinions are sharply divided over whether the agency can provide the kind of accuracy Ruckelshaus envisions.

"It's not a scientific process," one EPA scientist said. "Nobody faults it directly. It's really the best we can do. But when you go out for a beer after the meeting, they talk about how uncomfortable they are."

Inside the EPA, the overseer of risk is Elizabeth L. Anderson, director of the EPA's Office of Health and Environmental Assessment.

Anderson is in charge of the team that does the calculations, assessing the risks of perhaps 100 chemicals a year by selecting information or "data points" from laboratory results and plugging them into a complex mathematical formula.

The agency considers the assessments only a tool to help regulators do their job.

But risk assessment has become an increasingly important tool, one used as often to justify a decision not to regulate a pollutant as to justify regulating one. Not surprisingly, the assessments are often viewed with skepticism outside the agency.

"It's an inexact science at best, and all too often it's used as a rationale for not regulating," said Nicholas Freudenberg, a professor of public health at the City University of New York. "The fact is that there is a tremendous amount of uncertainty, and if you need certainty, even from animal tests, you're not going to have it for decades."

Like others at the EPA, Anderson harbors no illusions about the certainty of her office's product. Between the raw laboratory data and the finished product are dozens of assumptions, mathemetical manipulations and educated guesses.

"It's almost like a circumstantial murder trial," she said.

Another EPA official puts it more bluntly: "We have been accused of drawing numbers out of thin air -- and that may not be a bad analogy."

Consider the variables: A hypothetical chemical is tested in a laboratory on rats and mice of strains carefully bred to be extraordinarily sensitive to cancer-causing agents.

Male mice develop tumors at a high rate; female mice at a far lower rate. Rats develop no tumors at all.

Is a rat a better forecaster of human response than a mouse? Should the chemical be regarded as a human carcinogen? And if so, how powerful?

"We don't know which laboratory animals are closest to humans, so we have to take seriously the most sensitive animal model," Anderson said.

But human exposure to any chemical rarely comes close to the deliberately high doses administered to a laboratory animal. To complete the risk assessment, the EPA's mathematical model must have some estimate of real human exposure.

In most cases, hard numbers do not exist.

"We found out at Love Canal that we could spend $10 million at one shot and still not know what people are exposed to," Anderson said.

So the agency gobbles up whatever information it can find, often subjecting the raw figures to still more mathematical conversions.

If an agricultural chemical is used on apples, how much residue will remain on each apple? How many apples will a person eat each year? Will the residue remain if the apple is turned into juice, sauce or pastry? If so, how much juice will be drunk, how many pastries consumed?

What eventually emerges from all the estimates and calculations is a neat-looking figure that, according to EPA officials, is almost useless in predicting the incidence of cancer.

Critics contend that the process is nothing more than mathematics masquerading as science, and even the most devoted adherents of the art acknowledge that the risk figure suggests a mathematical certainty that does not exist.

If a risk assessment shows a 1-in-10,000 risk of cancer from a chemical to which 500,000 people are assumed to be exposed at certain assumed levels, for example, does that mean 50 cancers will result?

Maybe. But maybe not. Because of the conservative assumptions at each step of the process, the EPA says risk assessment produces an "upper bound" value. The presumed incidence of cancer will be no higher than 50, according to the EPA, but it could be considerably lower.

"I think that we're overstating the risks very substantially," one EPA scientist said. "I still think it's appropriate. I just don't think we should kid ourselves that this is an attack on cancer."

But other EPA officials privately worry that the assessments also may offer a false sense of security. The agency knows almost nothing about the possibility of synergistic or combined effects of exposures to many chemicals, for instance, and frequently is hampered by inadequate laboratory data, which forces still more assumptions, more manipulations, more guesswork.

"The answer eventually comes back that we don't know what to make of these models," Anderson said. "Given all of this, so much gets lost. You just see these numbers, and so many bodies."

Throughout the course of the day and night, the average American is exposed to hundreds of chemicals.

The Food and Drug Administration estimates that three-fourths of all foods contain minute residues of agricultural chemicals; processing may add others, in the form of preservatives, flavorings and colorings.

Inside the house, there are household cleaners, insect sprays, formaldehyde fumes. Freshly dry-cleaned clothes give off the faint aroma of perchlorethylene. A trip to the self-service gasoline station means breathing a little benzene, perhaps some ethylene dichloride, a few molecules of lead.

There's saccharin in the soda pop, chloroform in the air, trihalomethanes in the water.

Not all of the chemical substances in daily life are human creations. Cooking meat over high heat produces minute quantities of a cancer-causing substance. Aflatoxin, a carcinogen produced by a mold, may be found in nuts, peanut butter, even milk.

By statistical standards, none of those chemicals presents the risk that the average American assumes willingly by using the automobile. The risks of dying in an automobile accident in any given year are 1 in 5,000. Over a lifetime, the risk is far higher -- about 2 in 100, half that if seat belts are worn.

For some other self-imposed risks, the figures are even higher. For pack-a-day cigarette smokers, the risks of cancer are 1 in 10.

What, by contrast, are the risks from exposure to chemicals? Individual figures are hard to come by. According to the National Academy of Sciences, for example, fewer than 10 percent of this country's agricultural chemicals and 5 percent of its food additives have been fully tested for their ability to cause chronic health problems.

But consider the case of saccharin, one of the most exhaustively studied chemical compounds. According to a 1982 Harvard University study, drinking 40 diet sodas carries with it a one-in-a-million chance of cancer.

Someone who drank a diet soda each day for 70 years, then, would face a cancer risk of 6.4 in 10,000.

By one of the most commonly cited estimates, a 1981 study by two Oxford University scientists, about 8,000 Americans may die each year of cancer linked to exposure to environmental pollutants. The EPA has estimated that the annual number of cancer deaths linked to toxic air pollutants alone could range as high as 2,000.

The Oxford study estimated another 8,000 or so cancer deaths from food additives and "industrial products," perhaps 16,000 more from occupational exposures. By contrast, the same study estimated more than 120,000 annual cancer deaths associated with tobacco; 140,000 associated with diet.

Because a far higher proportion of deaths are attributable to tobacco and food habits, some health officials argue that the regulatory efforts on chemicals are entirely misplaced.

"If I ran the zoo," an EPA official said, "I'd outlaw smoking. You'd certainly save more lives."

But others worry that the impact of chemical exposures has yet to show up in the cancer statistics.

"When lung cancer became epidemic, it was a small proportion of total deaths," said epidemiologist Devra Lee Davis of the National Academy of Sciences. "We have no way of knowing until 40 years from now."

As Ruckelshaus tells it, the manager of a rubber-products plant had determined to level with his workers about the hazards of cancer-causing vinyl chloride and the new rules designed to limit their exposure to it.

"He went on at some length," Ruckelshaus said, "explaining the risks, and how safety rules would change, and health monitoring. At the end of the presentation, he asked if there were any questions, and there was a long silence. And I thought, boy, are they going to storm the stage or what?

"Finally, one worker piped up. He said, 'When are you gonna get this damn cigarette machine fixed?' "

Ruckelshaus tells the story to illustrate what some regulators see as the great paradox in risk regulation: The public will accept some very high and unquestionable risks, but will vehemently reject others of far lesser consequence.

In truth, neither the agency nor industry has had much success in relating risk to the public through comparisons -- the documented hazards of a high-fat diet, for instance, versus the less certain and presumably smaller risks of food preservatives.

"I used to think that had a lot of potential," Ruckelshaus said. "I'm not so sure anymore. They don't believe you, for one thing . . . . And it's been misused. People tend to compare everything to cigarettes, figuring that if the public accepts cigarettes as okay, then, Lord, anything's okay."

Critics of risk assessment, however, find no paradox in the public reaction. Reducing all risks to a set of finite and comparative figures, they argue, tends to obscure the fundamental difference between a voluntary risk and an involuntary one and ignores a person's right to conduct his own risk-benefit analysis.

A person who drives to work daily, for example, voluntarily accepts the hazards of the automobile in return for the obvious benefit: The car transports him to his job and back.

The same person may be unwilling to accept even the smallest amount of a chemical carcinogen that makes its way into his drinking water from a nearby hazardous-waste dump. Whatever the statistical risks involved, they far outweigh the nonexistent personal benefits.

A person in that position is not likely to be comforted by a regulatory calculation that finds the potential damage to his health is too low to justify the high cost of removing the chemical.

EPA officials confronted a similar reaction last year in Tacoma, Wash., where the agency conducted an unusual experiment in public education on the risks of arsenic from a copper smelter. The smelter provided a benefit to the community -- several hundred jobs -- but also a broadly shared risk of lung cancer from arsenic emissions that could not be totally stopped short of closing the facility.

The EPA held extensive hearings, trotting out its "upper bound" risk assessments. The smelter hired a public relations company to help explain the economic benefits of the plant. But by the end of the arduous experiment, dozens of Tacomans were sporting buttons that said, simply, "Both."

Some scientists and public policy experts rated the effort a failure, and have suggested privately that such official candor might not be the wisest course for the EPA to pursue.

Ruckelshaus flatly rejects that notion. "We've embarked on a process in which we will involve the public in this, and we can't back away," he said. "The more they know, the more comfortable they become with the concept."

As an example, he mentions the area of Washington state around Hanford, long the site of a disposal facility for low-level nuclear waste. The area has been designated as one of three possible locations for a high-level nuclear waste site, a choice that Ruckelshaus finds a logical one, at least in terms of public acceptance.

"They're comfortable with nuclear waste. They've lived with it for years," he said, adding wryly: "They may be wrong, but they're comfortable."