Since President Reagan took office, federal regulators repeatedly have refused to tighten restrictions on known cancer-causing substances in the air, water, work places and dumping grounds of America.

Many of these decisions are preliminary. But the Reagan regulators have denied almost all requests for immediate action, and have let existing proposals for action linger while they review the evidence. In effect, the administration has given chemicals the benefit of the doubt.

Among recent decisions about chemicals identified as carcinogenic, or cancer-causing, by the Health and Human Services Department's National Toxicology Program:

* The Environmental Protection Agency (EPA) refused to give priority status to regulation of formaldehyde as a toxic substance. Instead the agency has allocated $283,000 for a research information clearinghouse and a workshop of scientific experts. Industry officials challenge the assertion that formaldehyde is a human carcinogen.

About 28 million people living near chemical plants or petroleum refineries are exposed to low doses of it. The EPA is developing emission standards on synthetic organic chemicals, including formaldehyde.

* The Occupational Safety and Health Administration (OSHA) also refused a request for emergency regulation of formaldehyde, to which 1.8 million workers are exposed. The Consumer Product Safety Commission did vote to ban urea formaldehyde foam insulation.

* OSHA rejected petitions for emergency standards limiting exposure to the gasoline additive ethylene dibromide and the pesticide and sterilant ethylene oxide. A U.S. District Court judge this month ordered the agency to produce an emergency ethylene oxide standard; OSHA is appealing. An ethylene dibromide standard is being developed.

* OSHA ended preliminary work under way to develop standards for nickel and and cadmium, according to an agency official. The agency also is revising its overall policy on carcinogen regulation.

* The Food and Drug Administration proposed a new "constituents policy," which was developed during the Carter administration, allowing tiny amounts of cancer-causing chemicals to be added to food, drugs and cosmetics if they are an integral part of a substance which has not proved carcinogenic. Under this policy, the FDA has approved for use in cosmetics and drugs four dyes containing traces of carcinogens.

"People in a free society are at liberty to take certain risks. The regulator has to have a sense of what the society wants in terms of restrictions and fit the scientific evidence into the equation," said Arthur Upton, former director of the National Cancer Institute. "What worries me is the present administration seems to be going beyond a wholesome and reasonable reassessment of the evidence."

"Some people want to take a knee-jerk reaction. We want to be protective of the public health, but we don't want to fool the public and take an action just for the sake of saying 'I took an action,' " said John Todhunter, an EPA assistant administrator.

EPA did ban the carcinogenic pesticide toxaphene, although allowing use of existing stocks. Todhunter said the agency had kept more than 60 chemicals off the market under the chemical screening program.

Over the past decade, legislators, regulators and scientists have moved away from the rigid attitude exemplified by the 1958 Delaney Amendment to the Food, Drug and Cosmetic Act, which banned outright carcinogens in food additives.

Recent laws, such as the 1976 Toxic Substances Control Act, have required a more complex analysis balancing a chemical's risk, benefits and cost and the availability of technologies to reduce risk.

Recent government decisions have sharply accelerated the trend against new carcinogen controls, experts say.

There are sharp disagreements on the wisdom of this policy.

Ronald A. Lang, executive director of the American Industrial Health Council, which is supported by chemical manufacturers, said he applauds the decision to review what he described as the questionable scientific basis for some rules and proposals issued in the Carter administration's closing days. Lang also said he supports efforts like those of Administrator Anne M. Gorsuch to require more peer review of scientific studies.

But Tony Robbins, head of the American Public Health Association, and former head of the government's National Institute for Occupational Safety and Health, disagrees.

"Mostly, the [administration] is looking for a way not to act," he said " . . . We should accept a situation in which a small part of the time the tests proving carcinogenicity are wrong . . . to eliminate the situations in which people are not protected because a test showing no hazard was wrong."

For years, scientific regulators have tried to reconcile the legal precision expected in government rules with uncertainties inherent in science. Restrictions on suspected carcinogens often face challenges from all sides.

Court challenges tend to focus on a few key issues. How good is the evidence of carcinogenicity? How large are test samples? Are test animals predisposed to tumors of certain organs? And most importantly: if large doses of a chemical cause cancer in animals, how does one interpret the results to estimate if small doses cause cancer in humans?

Such questions are complex and emotionally charged. Projections show that one in five of the 2 million Americans who die each year, about 400,000, die from cancer.

The National Toxicology Program has identified 88 carcinogens based on its review of scientific evidence. Often the chemicals are key elements of a major industry; controls can cost millions of dollars.

In the late 1970s, the heads of the FDA, OSHA, the EPA and the CPSC appointed a staff to develop a concensus on some of these issues. In 1979, the group proposed the first government-wide cancer policy.

The proposal has not been revoked, but the White House Office of Science and Technology Policy has assembled a group of government scientists and regulators to prepare new guidelines for carcinogen research and regulation. Among other things, they are putting new emphasis on "risk assessment," a hybrid of statistics and science designed to show how many people are exposed to a chemical and whether exposure is likely to cause cancer.

A draft background paper circulated late last year by the White House technology office has been criticized by some experts who say they fear the findings could complicate carcinogen regulation. It drew praise, however, from the industrial health council, which objected to elements of the Carter administration policy, including the relatively heavy importance given test results that show carcinogenicity in animals and what the council called "negligible" weight assigned to human studies showing no risk.

Some critics expressed concern that the extensive discussion of the differing biological pathways chemicals follow in animals and humans could lead to de-emphasis of the kinds of animal studies used as a basis for much carcinogen regulation. There is also concern about the document's distinction between carcinogens that alter genetic material and those that promote uncontrolled division of cells.

This widely debated distinction could mean a more lenient approach to a chemical like saccharin, identified as a promoter of cancer but not implicated in genetic mutation. The FDA banned saccharin in 1977; Congress overturned the ban.