The nation's top experts on assessing whether chemicals cause cancer say their traditional methods are sometimes misleading and that improvements or entirely new methods should be developed.

In a three-day meeting that ended yesterday, the newly formed National Academy of Science panel on risk assessment said the use of rats and mice to test potential carcinogens -- a practice that has formed the basis for regulating chemicals in the United States for more than 25 years -- should be brought under sharp scrutiny. Many scientists said the studies are too unreliable and too inaccurate to form the basis for evaluating risks to humans.

The National Academy of Science panel, which was convened at the request of both the Food and Drug Administration and the Environmental Protection Agency, will conduct a series of meetings over the next 18 months as part of a broad-ranging review of the old methods.

Many scientists said they hoped the panel would arrive at a blueprint for radical reform, ending a system for calculating carcinogenic risks that they said has become outdated and that sometimes unfairly brands safe chemicals as dangerous and dangerous chemicals as safe.

"These {animal} tests can tell us whether a chemical has the potential to induce cancer in a laboratory animal under specific circumstances," said panel member Michael Gallo, a researcher at the University of Medicine and Dentistry of New Jersey. "They do no more than that. If we are going to use them for risk assessment, we need much more information."

The focus of the panel's first meeting was on the controversial practice of feeding test animals what is known as the maximum tolerated dose, or MTD, which is defined as the greatest amount of a given compound a rat or mouse can eat without being poisoned to death.

Scientists use the MTD for a simple reason. It is thought typical that a chemical might cause cancer in no more than one or two people out of a million exposed at realistic dose levels. That poses an enormous practical burden on scientists testing for cancer-causing potential. To spot that low risk in animals, they would have to give the realistic dose to at least 1 million rats or mice, a number that is considered much too costly.

Instead researchers feed rodents a much higher-than-normal dose -- in fact the highest dose possible, the MTD -- in order to cause more cancers than would occur at a more normal exposure, and then they estimate mathematically how many cases would be caused under the realistic conditions. Then they extrapolate to arrive at a smaller number thought to get cancer at a smaller dose.

In recent years many scientists have become increasingly suspicious of the value of this approach in light of new evidence that the tumors observed at the high dose may be as much the result of the size of the dose as of the nature of the chemical being tested.

At a workshop sponsored by the panel Thursday, University of California at Berkeley geneticst Bruce Ames -- a prominent risk analysis expert -- argued that when very large amounts of a chemical accumulate in an animal, they accelerate cell death and the proliferation of new cells -- a process that is known to increase the chance of a cell acquiring a cancer-causing mutation. Ames said some compounds that are safe at low doses may cause cancer at high doses, making the standard MTD extrapolation from high dose to low impossible.

Other testimony at the session suggested that high doses may alter the way in which a material is metabolized by the body or disrupt other physiological processes so as to make it dangerous when it would not be at lower doses.

Citing these concerns, many scientists said that it should no longer be sufficient for estimates of risk to be made on the basis of MTD results alone. Many scientists present said that before a finding of carcinogenicity from a test at the MTD is accepted, there should also be an explanation of why tumors appeared to be forming, showing that the culprit was not the dosage itself.

"Do these risk estimates at MTD have any real meaning? My opinion is that without some understanding of the {biological} mechanisms involved, they don't," said Thomas Starr, a scientist with Environ Corp., a consulting firm. "You need to know how it happened to make sense of high dose results."

Many scientists also said that regulators need to make it much clearer when dealing with the public or Congress that the results from animal cancer studies are far from definitive.

"I would prefer to keep using the high dose to see what the potential risk is, but to have the fortitude to say when that risk wasn't relevant in humans," said toxicologist Eugene McConnell. "But that is very hard to do. I have seen useful compounds prevented from reaching the market because animals got carcinogenic response that we knew had no risk to humans."