Inside the Military Entrance Processing Station in Linthicum Heights, Md., one day last week, a throng of shirtless young men waited to have their blood drawn. When a white-suited technician called out, "Next physical," another recruit would slide onto a stool and extend his arm. Deftly, the gloved technician tightened a tourniquet around the recruit's biceps and slipped a needle into a vein. Blood spurted into a vaccum-sealed tube.

The tube was spun in a centrifuge to separate the cells from the yellow fluid called serum, which was then poured into a numbered vial and shipped, along with dozens of other samples, to the laboratory that tests the blood of military applicants for exposure to the AIDS virus.

That procedure is repeated many times a day at the Maryland center and others like it around the country. Since it began almost two years ago, the Defense Department program has tested about 2 1/2 million people and has become the prototype for proposals to test engaged couples, hospital patients, immigrants and other groups. Its scientific standards are rigorous, the quality of the laboratory work is scrupulously monitored, and the results are impressive. Even so, no medical test is perfect.

Out of every 100,000 military applicants, about 150 test positive for infection with the AIDS virus. Between one and three of those 150 are "false positives," meaning that the tests indicate incorrectly that they have the AIDS virus. That estimate is based on using a variety of other laboratory tests to verify positive results and on following up samples that are equivocal with repeated tests.

"Everybody wants it to be a perfect test," said Dr. Donald Burke, chief of virus diseases at the Walter Reed Army Institute of Research. "That just ain't life."

The problem of false positives has received scant attention from advocates of widespread AIDS testing, but many public health experts predict it will have explosive implications if testing is expanded to include groups at very low risk of the infection.

"No one has asked the question, 'How many errors can we tolerate?' " said Dr. Stephen G. Pauker, a professor of medicine at Tufts University Medical School.

In an article published today in the New England Journal of Medicine, Pauker and Dr. Klemens B. Meyer, a colleague at New England Medical Center, predict that false positive rates will rise if lower-quality laboratories are used, if increased demand overburdens laboratories and if a new test is introduced as a substitute for the present, two-stage testing procedure.

Pauker said the false positive rate is critical in programs to test groups at very low risk of infection with the virus, such as marriage applicants in areas of the country with few AIDS cases or elderly patients scheduled to have elective surgery. In such groups, he said, the false positives could easily outnumber the true positives.

"The false positive rate will go up to some extent" as testing becomes more widespread, he said. "It doesn't need to go up very much to create a social catastrophe."

Current AIDS testing programs involve two tests, done sequentially. The first, the ELISA, is a simple screening test sensitive enough to detect almost all infected individuals, but it yields a high rate of false positives. If the ELISA is positive, it is usually repeated to verify the result. If the repeat is positive, then a different test, the Western blot, is used. This test has a much lower rate of false positives than the ELISA, but it is more complicated and expensive. Repeating the Western blot, trying to culture the virus and performing other blood tests can provide additional information but cannot prove with certainty that an individual result was a false positive.

"To prove absolutely that somebody is truly not infected, when all tests are positive -- there's just no way to do it," said Dr. Harvey Fineberg, dean of the Harvard School of Public Health.

The estimated false positive rate of one to three out of 100,000 is based on individuals who have gone through the whole testing sequence.

False positive tests make up a much larger proportion of all positive tests among people at low risk of contracting AIDS, such as couples about to be married, than in high-risk groups such as homosexual men or drug addicts, according to public health experts. For that reason, the decision to test such low-risk groups involves a tradeoff. To identify people infected with the AIDS virus, some uninfected individuals will be falsely labeled as carriers and will suffer the emotional and social consequences.

Because recipients of the false results would believe themselves infected, Pauker said, they might choose not to marry or have children. Their careers and insurance coverage could also be affected. In their article, Pauker and Meyer ask, "How many engagements should end to prevent one infection? How many jobs should be lost? How many insurance policies should be canceled or denied? How many fetuses should be aborted and how many couples should remain childless to avert the birth of one child with AIDS?"

Burke argues that the track record of the military program shows that widespread testing can nevertheless be successful if done very carefully.

"Critics of testing have said {it's} logistically unfeasible, economically unsound and inaccurate. All of those are untrue," he said.

But he added, "I'm not sure that we are nationally ready to implement screening programs. It's not because we can't. It's because we haven't devoted the attention to make damn sure that the tests work very well" at all the laboratories involved.

Both the ELISA and the Western blot measure antibodies against the AIDS virus present in the blood. Such antibodies usually develop between three and 12 weeks after an individual becomes infected with the virus.

Burke said the quality of the tests, particularly the Western blot, varies greatly among laboratories. The Western blot's false positive rate depends both on how well it is done and on the strictness of criteria for calling a test positive. The test can show a pattern of one or more bands or stripes corresponding to antibodies against different proteins in the human immunodeficiency virus (HIV), the cause of AIDS.

Some laboratories consider a test positive if just one band is visible, which increases the likelihood of false positives. Burke said the military program is about to change its criteria to require the presence of two bands. The American Red Cross requires three bands, according to Dr. James AuBuchon, a medical officer there.

Based on research in the military program, Burke estimates that the proportion of positives that are false is about 2 percent, according to Burke.

Thus, if a military applicant tests positive, "there is a 97 to 98 percent chance that they really are infected," he said.

But some public health experts, including Pauker and Fineberg, contend that even a false positive rate of three in 100,000 could be a major problem if the population being tested had a very low incidence of infection with the AIDS virus.

For example, American Red Cross statistics show that 10 of every 100,000 potential blood donors test positive for the AIDS virus. Donors who test positive are notified, and their blood is rejected. If the false positive rate were three out of 100,000, it would mean that three of those 10 positive tests, or 30 percent, were in error.

AuBuchon said the American Red Cross program considers all its positive Western blots to be true positives. "There aren't false positives," he said. He said he was unaware of any studies to the contrary.

Dr. Harold Jaffe, chief of AIDS epidemiology at the Centers for Disease Control, said the success of the military screening program had persuaded CDC officials that the false positive rate can be kept acceptably low. "I think our general feeling is, it's not really that much of an issue," he said.

At the military processing processing center in Linthicum Heights, Capt. Michele Welles, the Army medical administrative officer, said that if a blood sample tests positive on two ELISA tests and a Western blot, the applicant is notified and the Western blot repeated on a second sample. She said that if that is also positive, the applicant is disqualified from military service. She said a doctor at the station talks with such applicants individually, advising them to see a civlian doctor for further evaluation.

Burke said he is pleased with the results of the military program but acknowledged that similarly high standards might not be achieved by other testing programs. He said that the problem of false positives in AIDS testing is no different from those encountered with other diagnostic tests.

"What it means is, if we are going to go ahead we're going to have to make sure that we do it right," he said.

The AIDS virus contains proteins. If a person becomes infected with the virus, his immune system starts manufacturing antibodies to tag these proteins. The process takes three to 12 weeks.

The two test procedures described below involve measuring the presence of antibodies in the blood.

The AIDS virus contains proteins. If a person becomes infected with the virus, his immune system starts manufacturing antibodies to tag these proteins. The process takes three to 12 weeks.

Two of the test procedures described below involve measuring the presence of antibodies in the blood. The third, very different process is not routinely performed at testing centers.

TESTING FOR THE AIDS VIRUS

THE ELISA TEST

(Enzyme-Linked Immunosorbent Assay; the most common screening test for infection)

Proteins from the

AIDS virus are spread onto tiny plates or beads.

A sample of the blood

to be tested is poured into small wells containing the protein-coated plates/beads.

If antibodies are

present in the blood, they stick to the proteins.

A detecting antibody

is then added to the wells. It will bind to any antibodies stuck to the proteins and cause a change of color. (Based on the color, the laboratory reports positive or negative test results.)

THE WESTERN BLOT (Used to confirm a positive ELISA test)

Proteins from the

AIDS virus are separated in a gel and blotted onto special paper.

A sample of the blood

to be tested is applied to the paper.

If antibodies are

present in the blood, they stick to the proteins.

A detecting antibody

(radioactive or chemically treated) is then added to the paper. It will bind to any antibodies stuck to the proteins and cause a pattern of one or more bands to appear. (Based on the number and position of the bands, a technician reports positive or negative test results.)