It's a question at the heart of the animal rights debate: Can drug or cosmetic companies studying a new product's toxicity substitute test tubes and computer programs for rabbits and rats?

Many opponents of animal research say yes. Scientists who work with new methods -- including those who developed the tests -- say such tests can replace some animal testing now and that in coming years new tests will replace more animal testing. But they say it may never be possible to replace animals completely.

While improved technology and increased public awareness have brought what some call a revolution in the field, toxicologists say these tests give much less insight into the overall effect of a new drug or cosmetic on the body than does giving the substance to a whole, living animal.

"Ten years ago nobody was looking into {nonanimal} toxicity testing because they thought that the mechanisms were too complex," said Dennis Stark, an associate professor at New York's Rockefeller University who has worked on nonanimal toxicity tests. "Today we've developed a lot of those assays and found that they're good for some things and not good for others. No one test is going to give the answer for all of these compounds and that's what's scary."

But many toxicologists say the tests are useful in helping weed out chemicals so they don't have to waste time doing animal tests on some substances. The tests also help scientists focus on certain effects of a substance when they do the animal tests.

"We're not developing alternatives, we're developing adjuncts," said Melvin Andersen, head of risk assessment at the Chemical Industry Institute of Toxicology in Research Triangle Park, N.C. "The goal of the adjunct is to help reduce the number of animals we use."

Toxicologists give different reasons for why animal tests cannot be eliminated. Many of those who work with tissue cultures say they are able to observe the effect of a chemical on all of the body's organs, but they say their techniques are not replacements because it is impossible to figure out the concentration of most substances to which different organs in the body are exposed. That concentration depends on a complex array of factors such as how much enters the body through the skin or the lungs or the intestine, how much is broken down by the body's detoxifying systems in the liver and how much simply passes through the body to be excreted. The long-term effect also depends on how much may be sequestered in different organs or in fat and muscle.

Computer specialists believe they can figure out this concentration. But they argue that individual tissue culture systems do not provide enough information about how a chemical will act in the body.

"We don't have a way to study all of the cells and to know what is the critical target organ," Andersen said. "The integration of what happens in an animal is not just what happens in a cell."

One of the first nonanimal toxicity tests that was widely accepted was the Ames test, a quick method of screening chemicals for their ability to cause mutations and, possibly, cancer. The test involves growing special bacteria on a type of medium deficient in a critical nutrient and observing whether adding the chemical in question makes them mutate, thus allowing them to survive.

The test has become a standard for companies screening new substances. But even though it is considered a good screen for mutagens, many of which cause cancer, it cannot pick up substances that cause cancer by other methods.

Like many such tests, the Ames is usually used before a researcher decides whether to test a substance on animals. If the compound is found to be highly mutagenic, researchers often can rule out testing it on animals -- thus saving time and money, as well as sparing the animals. An Ames test costs about $500 and takes an afternoon, while an animal cancer test can take a couple of years and cost more than $1 million.

"It's not a replacement," said Bruce Ames, the University of California at Berkeley scientist who developed the test.

But a substance's cancer-causing potential is only one of the things toxicologists look for when evaluating a substance. They also have to determine whether it will be toxic by directly killing or injuring cells.

Until 10 years ago, this was nearly impossible outside of whole-animal tests, partly because scientists lacked a means of growing cells in a way that mimicked the way they grow in the body and partly because they had a much more limited understanding of the ways substances might affect cells.

One new technique involves a type of culture system that has been marketed within the past five years. At Bristol-Myers Squibb, the system has enabled toxicologists to grow certain specialized kidney cells in a continuous layer, one cell thick, much as they live in the kidney. Researchers are especially interested in observing different chemicals' effects on such cells because kidney cells are heavily exposed to toxins. That's because the kidney concentrates materials from the blood, including toxins such as drugs or drug breakdown products.

Using this system, toxicologists can see whether a substance affects the cells' transport of molecules such as sodium and glucose, two compounds they normally carry across their membranes. If the transport is impaired, they can conclude that the substance injured the cells.

Other types of cells also can be grown using this system. But as with the Ames test, toxicologists say this test is mainly of use in screening substances to rule out ones that are not worth testing on animals.

Many researchers say such screening tests have resulted in a drastic reduction in the numbers of animals used for testing. Whether that is true is difficult to determine, since no specific figures exist. In 1988, according to the Agriculture Department, 1,635,288 animals were used in the United States for experimentation -- but that includes basic science research, as well as routine testing.

Alternatives also have been developed for the so-called Draize tests, in which chemicals are placed on rabbits' eyes or skin to see whether they cause irritation. The Draize tests have been widely criticized by antivivisectionists. The alternative tests, which use such things as egg membranes and cell cultures, are generally designed to measure how substances affect the cells of the cornea, the clear part of the eye.

Some scientists have criticized the alternatives to the Draize because such tests are done outside a body and thus cannot measure a reaction such as inflammation, in which certain types of cells travel from elsewhere in the body and attack the eye. But others say that the Draize isn't a good measure of inflammation either, since the test is scored in a way that doesn't weigh this reaction as heavily as others.

The controversy over alternatives to the Draize, many scientists say, illustrates the need for some generally accepted organization that would analyze all of these tests objectively. Getting regulatory agencies such as the Food and Drug Administration to accept nonanimal test results when evaluating a new product can be slow, some say, because many scientists are skeptical about the accuracy of these tests. Others say companies have been slow to adopt nonanimal tests because they fear lawsuits, since someone might have a bad reaction to a product and charge that it was not adequately tested. If such tests were validated by an independent panel, they say, more companies might be willing to use them.

"The technology has leaped ahead," said Andrew Rowan, director of the Tufts Center for Animals and Public Policy. "In that sense there's been a revolution. But we haven't yet had that final step in the policy arena."

Others say such a mechanism isn't needed, since techniques that are good science, such as the Ames test, will inevitably gain acceptance. They say organizations such as the Johns Hopkins Center for Alternatives to Animals, which was started 10 years ago largely with funding from the cosmetics industry and sponsors annual symposia on nonanimal testing methods, and the publication of four new journals devoted to nonanimal testing methods will help speed acceptance of good tests.

"There is every reason to believe that the normal processes of scientific communication -- journals, peer review and so forth -- will do a great deal to justify our use of these tests," said Oliver Flint, a research toxicologist at Bristol-Myers Squibb. "The FDA consists of scientists just like us."

Even if the tests could replace ones done on whole animals, they would not completely eliminate the use of animals. The kidney tubule cell model, for example, uses cells from a rabbit kidney.

"Most {of these tests} are reducing the animal's pain and suffering," Rowan said. "That's a good place to start."

Animal rights advocates respond that the animal tests have major limitations and say the public does not need many of the products being tested.

"I'm willing to concede that certain animal tests are very difficult to replace," said Neal Barnard, president of the Physicians' Committee for Responsible Medicine, an antivivisection group. "But the animal tests are highly imperfect and they're not getting any better. We have to invest in alternatives."

He added that many tests could be eliminated altogether if the FDA required drugs to be "necessary," as well as "safe and effective" -- thus eliminating testing on drugs that he says are not needed.