The delay in confirming the United States' second case of mad cow disease seems to underscore what critics of the agency have said for a long time: that there are serious and systemic problems in the way the Agriculture Department tests animals for mad cow.

Indeed, the lengthy delay occurred despite the intense national interest in the disease and the fact that many countries have banned shipments of beef from the United States because of what they consider to be lax testing policies.

Until Friday, it was not public knowledge that an "experimental" test had been performed last November by an Agriculture Department laboratory on the brain of a cow suspected of having mad cow disease, and that the test had come up positive.

For seven months, all that was known was that a test on the same cow done at the same laboratory at roughly the same time had come up negative. The negative result was obtained using a test that the Agriculture Department refers to as its "gold standard."

The explanation that the department gave late Friday, when the positive test result came to light, was that there was no bad intention or cover-up, and that the test in question was only experimental.

"The laboratory folks just never mentioned it to anyone higher up," said Ed Loyd, an Agriculture Department spokesman. "They didn't know if it was valid or not, so they didn't report it."

On hearing that Friday night, Dr. Michael K. Hansen, a senior research associate at Consumers Union and frequent department critic, reacted skeptically.

"That seems hard to fathom," he said. "If it's true, we have a serious communication problem at the Department of Agriculture. How can we be confident of anything they're saying?"

Mr. Loyd, reacting to a reporter's question about the Agriculture Department's handling of the issue, said, "In hindsight, reporting it would have been the thing to do."

Agriculture Secretary Mike Johanns briefly mentioned the positive test result at a news conference on Friday. The primary focus of the conference was to announce that British scientists had confirmed the United States' second case of the disease.

The sequence of events started in November, when an Agriculture Department laboratory in Ames, Iowa, performed two tests on the animal in question. After the "gold standard" test came up negative, the agency announced that the animal had not had mad cow disease. But at the same time, the same lab also conducted the experimental test, with different results.

Then two weeks ago, for reasons that are unclear, Phyllis K. Fong, the Agriculture Department's inspector general, arranged for further tests on specimens of the same cow. A test known as the Western blot, which is widely used in England and Japan but not in the United States, came up positive.

Because this result conflicted with the "gold standard" result from November, a specimen from the same animal was sent to a laboratory in Weybridge, England, that is considered pre-eminent in its field. Several tests were conducted there, and all of them came up positive; it was the results of those tests that Mr. Johanns announced at the news conference on Friday afternoon.

The nation's mad cow testing system is now infuriating both ranchers and consumers. Consumer lobbyists say the flawed results show once again that 15 years of testing has been dangerously inadequate. And now the beef lobby, which has long enjoyed a cozy relationship with the Agriculture Department, is complaining that the testing system is dangerously unpredictable.

Jim McAdams, president of the 25,000-member National Cattlemen's Beef Association, has complained that unexpected testing creates "great anxiety within our industry," and leads to "significant losses."

Thirty-six countries have shut their doors to American beef, virtually wiping out a $3 billion export market, which Australia happily moved into.

On Saturday, Taiwan reimposed a ban on American beef that it had lifted just two months ago, Reuters reported.

The new case of mad cow appeared to be in a native-born animal, though Mr. Johanns was vague about that. Testing also suggested that the animal caught the disease from a new food source, since the strain was different from that of the Washington State cow that tested positive in 2003.

Mr. Johanns said that catching one positive in 388,000 recent tests proved the system worked.

But critics said it did no such thing, because the system was designed strictly for surveillance. One positive caught after a seven-month delay was, at best, a stroke of luck, the critics said.

Other countries use food-safety standards: Japan tests every cow, Europe tests about one in four.

The United States instead uses statistical models that it says will let a few tests detect the infection even in one cow in a million. It now tests one in 90; when the first mad cow case was found in 2003, it was testing one in 1,700.

With its statistical logic under regular attack, the United States has increased the number of tests to 388,000 in the past year, from 40 in 1990. But until recently, Mr. Johanns was discussing cutting back to 40,000 tests.

That system is "bizarre, illogical and woefully inadequate," said John Stauber, co-author of the book, "Mad Cow U.S.A.," which was first published in 1997.

"The bottom line," he said, "is that the U.S. government is afraid of putting in real food-safety testing because it would certainly find additional cases."

Mr. Loyd of the Agriculture Department replied that surveillance testing assumed a few animals would be positive, and that his department had nearly doubled its own goal of testing 220,000 cases in a year.

"There is no scientific basis," he said, for doing what Japan and many critics want: testing all animals or all those more than 20 months old.

But even a scientist who helped design the department's testing now harbors doubts about it.

In an interview before the second case was found, Dr. Linda A. Detwiler, who retired in 2002 as the chief of the mad cow testing program and now teaches veterinary medicine at the University of Maryland, said the department should be using the Western blot test it was resisting.

"You need to put as many tools in your tool kit as possible," she said.

Mr. Loyd said Secretary Johanns now agreed. The beef industry now cites its consumer-protecting "firewalls."

But it took many years to erect them: a ban on feeding ruminants to cattle, a ban on using near-dead dairy cows as beef and a ban on using the brains and spinal cords of older cattle in feed.

Other practices that many veterinarians dislike continue, such as feeding poultry litter with spilled cattle meal in it back to cattle, giving calves "milk replacer" made from cattle blood and letting cows eat dried restaurant "plate waste."

Dr. Detwiler was adamant that those practices should end, and that the brains and spines of all cattle should be destroyed, not made into feed even for pigs or chickens.

"That's how you keep infectivity out of the food chain," she said. "If a farmer makes a mistake and gives pig feed to cattle by mistake, the feed is safe."

The beef lobby has opposed many changes, and statements from the industry and the lobby often echo each other. When Mr. Johanns held a sort of pep rally for beef in Minnesota recently, no consumer groups were on a panel that declared American beef "very, very safe," but lobbyists were.

The industry casts a long shadow over the department. Ann M. Veneman, a former agriculture secretary, had as her spokeswoman Alisa Harrison, who, in 1996, accused a doctors' group of being an animal rights group opposed to eating meat. The doctors' group had endorsed the ban on feeding cattle or sheep to cattle.

Mr. Johanns, a former Nebraska governor who grew up on a dairy farm, inherited two officials of the cattlemen's group, Charles Lambert and Dale Moore, as deputies. His under secretary for farm and foreign agriculture services, J. B. Penn, worked at a consulting firm serving the industry.

Last week, according to the Kyodo news service in Japan, a group of Japanese lawmakers who visited Mr. Penn in Washington accused him of "threatening" them with trade retaliation and saying that the United States' patience was growing short and that they should simply accept American beef.

Mr. Loyd denied that, saying the meeting was "cordial."

Such connections to industry impede the department's duty to police it, said Representative Rosa L. DeLauro, a Connecticut Democrat on the House Appropriations Committee. (On Tuesday, the department announced $140 million in grants to advertise American food overseas, including $12 million to the U.S. Meat Export Federation.)

She wants a new, separate food safety agency, like the one Britain created in 1999.

But the department's harshest critic has emerged from within - it is Ms. Fong, the inspector general, who ordered the new round of tests.

Last spring, she issued a scathing analysis of the testing program, saying, for example, that it could not do scientific sampling because it was voluntary, tested too few healthy-looking cattle, and could not assure that sick but walking cattle and cattle that died on farms were tested.