The Agriculture Department is defending its decision to release results of preliminary tests that raise concern about a possible mad cow disease (search) infection when the initial findings may well be wrong.

The announcement last Friday that a cow carcass had not passed a preliminary screening test for mad cow caused concern on commodity markets, worried consumers and angered some farmers. Assurances by the government that the initial screening was only designed to trigger more sophisticated tests did little to ease the anxiety.

On Wednesday, five days after the initial announcement, the Agriculture Department confirmed the carcass in question did not have mad cow. Sophisticated chemical tests conducted by the USDA's National Veterinary Services Laboratory in Ames, Iowa, came up negative for the disease.

The lab is continuing tests on a second carcass that failed an initial screening on Tuesday, with confirming results not expected for several days. In both cases the department refused to provide any details about where the cows were slaughtered, where they were screened or their age - only that none of the meat had been allowed into the food chain.

Mad cow disease is also known as bovine spongiform encephalopathy (search), or BSE. People who eat products containing the BSE protein can contract variant Creutzfeldt-Jakob (searchdisease, a rare but fatal disease similar to mad cow.

Friday's disclosure raised questions about why the department had decided to announce anything at all, since the initial screening is known to cast a wide net that is bound to include "false positives" that are later discounted.

"USDA wants to be very transparent with this issue," said John Clifford, deputy administrator of the department's Animal and Plant Health Inspection Service and the USDA's chief veterinarian.

He acknowledged in a conference call with reporters Wednesday that there had been "a lot of discussion" on whether to make public the initial screening results. But, he added, "we realized that information like that may be leaked. And we want to be very open and transparent, and that's why the decision was made to release this type of information."

Clifford repeatedly has said the screening tests, which are extremely sensitive, are not meant to show presence of mad cow disease, which attacks a cow's nervous system and brain. They are only expected to disclose the potential of an infection in a high-risk animal.

The only case of mad cow disease in the United States was discovered in a Canadian-born Holstein on a farm in Mabton, Wash., in December. That prompted the Agriculture Department to step up its screening of animals, especially "downers" that are unable to move or animals whose cause of death is unknown.

Since June 1, when the new screening began, 8,587 cows have received the preliminary test. Two have showed an "inconclusive" result, meaning a follow-up test was needed.

While the USDA expects a certain number of cows will be singled out in the screening, it has not determined how many of those actually are likely to be infected - or how many will be "false positives."

The mad cow screening programs used by the USDA - and developed by Bio-Rad Laboratories of Hercules, Calif. — have been used in Europe for a number of years.

According to the World Organization for Animal Health, 5.5 million cattle were tested in 2003. The screening found 119 cases of possible infection - "inconclusive," in the U.S. vernacular — with 85 percent of those later confirmed to have mad cow disease. That amounts to 18 false positives as a result of the screening.

Clifford said he could not give any estimate on the rate of false positives in the USDA screening program.

"We prefer not to put any type of statistic on that," he said. "You can look at some of these tests and their use internationally, but I think we need to have that experience here with the U.S. and the population of animals that we're testing before we prefer to give any type of statistical number."

The department plans to screen 268,000 animals over the next 12 to 18 months. Agriculture Undersecretary J.B. Penn recently said he anticipated one "inconclusive" result in every 10,000 cattle screened. But Clifford wasn't comfortable with that estimate, saying it was based on Japanese test results using different classes of animals.

"We could see results that differ from what they're seeing," said Clifford.