Doctors should think again about its use, say the authors. "A positive HIV status has such profound implications that nobody should be required to bear this burden without solid guarantees of the verity of the test and its interpretation," they conclude.

The findings, likely to cause intense debate in the medical fraternity and anguish for many HIV-positive people, are contained in an article published by the respected science journal, BioTechnology.

Many people who appear to be infected with HIV, say the researchers, can be suffering from other conditions such as malaria or malnutrition that produce a positive result in the test. Even flu jabs can produce the same effect. As a result, predictions by the World Health Organisation (WHO) that millions are set to die because of being HIV-positive may be wildly inaccurate.

The paper also lends powerful support to the theory, held by growing numbers of scientists, that HIV is not the true cause of AIDS. One of its authors, Eleni Eleopulous, a biophysicist at the Royal Perth Hospital, said this weekend: "There is no proof that people labelled as 'HIV-positive' are infected with such a retrovirus. We should really question the role of 'HIV' in the causation of AIDS."

Overall, the findings "mean the tests have not been scientifically evaluated", she said.

The authors say that neither of the two main HIV tests used have been adequately checked for accuracy. These tests rely on detecting antibodies to HIV in blood samples. But people whose immune systems have been activated by several other conditions, including tuberculosis and multiple sclerosis, can trigger the same reaction, giving a false-positive result.

Promiscuous homosexual men, illicit drug users, multiple blood transfusion recipients such as haemophiliacs and people subject to multiple infections become increasingly liable to give a positive result the longer their immune system is weakened, regardless of HIV.

To have confidence in antibody tests, they must first be validated by having their results checked against a "gold standard" that is, isolation of the virus itself. However, this has never been done with the AIDS test. The report adds that a procedure used to confirm the validity of diagnostic tests by looking for a virus's genetic material has also been shown to produce false results and cannot be considered as synonymous with isolating the virus.

The AIDS tests look for the detection of a protein called p24, generally considered the equivalent of isolating the virus. However, it has been detected in one out of 150 healthy individuals, 13% of people suffering from warts a condition that signals a weakened immune system and 41% of patients with multiple sclerosis, another immune system disorder.

Heavy exposure to sperm can also set up an antibody reaction, especially when entering the body through anal intercourse. It is another probable source of false-positives.

The WHO, which is seeking an extra Pounds 2billion a year for its AIDS prevention programme, estimates that about 14m people have been infected with HIV worldwide. It claims the total will reach 30-40m by the year 2,000, and that most will eventually contract AIDS.

Developing countries are said to face the biggest threat, with Africa alone already having an estimated 8m HIV-infected people. However, according to the BioTechnology report, these are the countries where the tests may be at their most unreliable because of widespread ill-health caused by other diseases. Severe malnutrition and multiple infections are especially likely to produce a misleading result in the test. Claims that current AIDS tests are virtually 100% accurate are based on studies of healthy subjects.

Eleopulous said that the paper, which underwent detailed scrutiny by other experts, concentrates on the shortcomings of one of two main categories of "AIDS test" known as "western blot", generally considered the more definitive of the two.

However, she said doubts were even stronger over the validity of the other test, called Elisa. This is usually administered first but is widely acknowledged to carry a high risk of false-positives.

Screening with Elisa in Russia in 1991 produced 30,000 positive tests, of which only 66 were confirmed using western blot. In the United States, a study among military applicants produced 6,000 individuals with an initially positive but subsequently negative Elisa test.

Dr Philip Mortimer, of the virus reference division at Britain's Public Health Laboratory Service, accepted last week that some fair points about the weakness of the western blot had been made, but he rejected claims that Elisa was even worse. He maintained that the situation described in the article was not typical of this country, where there is less reliance on western blot.

An initial positive test would be followed by a combination of different Elisa tests, although sometimes including western blot, and a test of a follow-up specimen, said Mortimer. "Only if the positive reactions on both specimens are confirmed, usually in a reference laboratory, is a positive report issued."

He believed there was no evidence that people had been falsely told in Britain that they were HIV-positive.

The findings have been welcomed by Professor Peter Duesberg, a top American virologist who maintains that HIV is not the cause of AIDS. He said the evidence helped to explain how "a false correlation" had been found between "HIV" antibodies and AIDS.

"The whole virus hypothesis of AIDS is based on this correlation," he said. "Its proponents have nothing else: no mechanism whereby HIV could do the damage attributed to it, no animal tests, no cure, no vaccine, no virus activity.

"They have nothing conventional in terms of virus-disease argument, except this correlation with antibodies. If this study is correct, and I have no reason to doubt it, it means that even that is now falling apart."

The findings have already led to a call by the New York Native, an influential gay newspaper, for legal action against the American government by relatives of people who have killed themselves, or suffered toxic effects from taking the anti-viral drug AZT, as a result of positive HIV tests.

Charles Ortleb, the editor, said: "If the test doesn't work, and if people really don't know that they are infected, the whole thing needs to be rethought ... This should be given high priority by the research establishment. We think that as a practical matter, no one should trust this test." *