(FORTUNE Magazine) – A decade ago, in a publicity stunt to convince the world that mad cow disease couldn't infect humans, British Agriculture Minister John Gummer fed his 4-year-old daughter a hamburger on the steps of Parliament. Later it became clear that the message in that photo op was terribly wrong. The girl never got sick, but dozens of other people did, succumbing to an invariably fatal disease that riddled parts of their brains with microscopic holes. The ensuing mad cow crisis has cost Britain alone some $7.5 billion and prompted the slaughter of more than 175,000 cattle. It has killed about 100 people and potentially doomed tens of thousands more who may be infected but haven't yet shown symptoms. Around the world, mad cow has evolved into a menace that just won't die--as soon as the threat wanes in one place, it pops up in another.

The latest uproar was in northern Vermont, where two flocks of imported sheep--one in the Mad River Valley, of all places--were suspected of harboring the disease. This was America's first skirmish with mad cow, and the U.S. Department of Agriculture took pains to show it isn't about to follow Mr. Gummer into the annals of ignominy. Though there was only circumstantial evidence that the sheep posed a serious risk, the agency seized them in late March and trucked them off to slaughter.

Critics assert that the USDA overreacted, pressured by a $50-billion-a-year American beef industry anxious to protect its valuable mad cow-free status. No one can deny that if just one mad cow is found in the U.S., beef producers will see demand drop faster than you can say "tofu." But what happened in Vermont was more complex than that. Like bear markets, mad cow affairs are all about uncertainty, which allows worst-case scenarios to loom large, precipitating drastic acts. In seizing the sheep, the USDA acted with the same kind of anxious intensity that investors have shown selling stocks during recent bouts of "mad Dow."

Much of the toxic uncertainty about prion diseases--the family of brain-spongifying ailments that includes mad cow--stems from the fact that creatures can harbor infections for years without showing symptoms, giving the disease a chance to spread. The USDA was especially worried about that in Vermont. Studies suggest that if mad cow disease jumped to sheep, it would be even harder to contain than it has been in cattle--a concern that tended to get lost in TV sound bites about the heart-rending story.

Fretful conjecture also swirls around federal efforts to keep mad cow off American soil. In January the Food and Drug Administration reported that 9% of FDA-licensed feed mills it surveyed did not have systems in place to prevent mix-ups that could put protein from cows and other cud-chewing animals into cattle feed--such cannibalistic feed, banned in the U.S. in 1997, is probably what spread mad cow disease in Britain.

But despite the Stephen King moments you've probably had biting into burgers, mad cow disease is generally thought to pose little risk in the U.S. If you're determined to worry about food, consider this: Listeria monocytogenes, a food-borne bacterium that often infects pregnant women, kills about 500 Americans a year. By comparison, mad cow has killed about 20 people a year since 1996, almost all in Britain.

Even taking mad cow's long latency period into account, the threat seems small compared with scourges like hepatitis C and AIDS. Epidemiologists think that tens of thousands of people may be infected with mad cow. But as the years go by without large numbers getting sick, estimates of the maximum number of infected people have come down--from many hundreds of thousands to around 100,000. Due to the latency period, experts say they'll still need five years or more to get a real handle on the probable toll. Meanwhile, thanks to food-safety measures, new exposures to the disease are thought to be virtually nil in most places; mad cow is now considered a nonthreat even for British beef eaters. (The more recent outcry about U.K. livestock concerns foot-and-mouth, a viral disease that doesn't pose a significant health risk to people. For a traveler's advisory on mad cow risks worldwide, see www.cdc.gov/travel/madcow.htm.)

Still, the U.S. has its guard up. To keep American livestock mad cow-free, the government maintains a swat team poised to act fast if a brain-addled bossy turns up. The USDA has an emergency plan that specifies invoking quarantines, assembling staffers in a situation room near Washington, D.C., and dispatching an official to hand-carry cow-brain samples to Britain for a second opinion. Part of that swat team swung into action in Vermont, where the seizing of the sheep offered a revealing window into the anti-mad cow war.

The USDA command center in Montpelier, Vt., was easy to find last month. Outside the federal building, a stone structure in the middle of town, protesters waved signs with messages like BEWARE! THE U.S. KILLS HEALTHY SHEEP FOR MAD COW BUT DOES NOT TEST ITS COWS. Inside, the task force leader, USDA veterinarian Linda Detwiler, strategized with colleagues in a makeshift situation room replete with aerial photos of one of the farms involved, dotted with little pink arrows labeled SHEEP.

Detwiler is the USDA's go-to person on mad cow issues. Tough, terse, and all business, she grew up on what she calls a "garbage-eating-hog farm" in New Jersey. She has been dealing with spongiform disease since the 1980s, when she led the USDA's effort to eradicate scrapie, an insidious sheep sickness that, like mad cow, fills animal brains with holes. "I saw grown men cry" as their sheep were led away, she grimly recalls.

The trouble in Vermont started in 1998, when the USDA learned there was a small chance that two flocks of sheep, or their progenitors, had eaten feed containing cattle protein before they were imported from Belgium. Larry and Linda Faillace of Warren, Vt., had imported the rare East Friesian sheep on behalf of Houghton Freeman, a Stowe, Vt., philanthropist whose father co-founded the global insurance giant American International Group. Freeman, 79, aimed to sell the flock's offspring from a Vermont farm he owns in order to help rejuvenate small farming in New England--East Friesians produce scads more milk than most sheep and are prized by makers of gourmet cheese. The Faillaces set up a similar breeding operation at their farm.

The couple contended that records from Belgium showed the sheep never got the risky feed. But the USDA wasn't convinced, for European controls on feed had failed to prevent Britain's mad cow epidemic from spreading to the Continent. In late 1998 the department quarantined both Vermont flocks, putting the owners' breeding plans in limbo.

Detwiler also was worried by fragments of evidence suggesting that diseased sheep may pose a greater public-health risk than diseased cattle. Sheep, for some reason, seem quite susceptible to mad cow: A 1993 British study showed that sheep in the lab got a form of the disease when fed only tiny amounts of brain extracts from infected cattle.

That susceptibility increased the urgency of a long-standing public-health question: If spongiform disease can leap from cows to people, can it also leap from sheep to people? The answer is murky, as is much about spongiform diseases. In fact, scientists still haven't nailed down what prions, the molecular entities that cause the diseases, exactly are. According to one theory, prions are perversely twisted proteins that act like Typhoid Mary crossed with Hannibal Lecter. Merely by touching certain normal proteins within the body, prions turn them into spitting images of themselves, with the same malign power of transformation. As they replicate, the rogue proteins eat holes in the brain, giving it a spongy look under the microscope.

But some data are hard to reconcile with this idea, including the baffling discovery that prions are still able to infect animals after exposure to furnacelike temperatures that turn proteins to ash. There's also the bizarre fact that prion diseases can be both contagious and inherited. The infection route was elucidated in the 1960s when researchers showed that women of New Guinea's Fore tribe contracted a spongiform disease by carving up and eating their deceased kin who had died of the disease. Dubbed "laughing death," the malady mocked its victims by afflicting them with an uncanny semblance of the giggles. The genetic route, meanwhile, has been shown to cause fatal familial insomnia, a human spongiform disease that blocks sleep until its victims become demented, go into a coma, and die.

Detwiler was also troubled by evidence that an outbreak of mad cow in sheep would be hard to stamp out. Cows don't transmit mad cow very easily: Disease prions are mainly found in the animals' brains and spinal cords, so the disease in cattle probably spreads only when they're given feed containing nervous-system tissues from infected animals. But mad cow prions in sheep seem more likely to spread: A 1996 study of experimentally infected sheep showed that the prions become widely disseminated in the animals' bodies. That means the sickness might easily travel through flocks by routine contact among animals--a pattern Detwiler knew well from her years battling scrapie, the spongiform disease that is most common among sheep. "Presymptomatic" ewes typically pass scrapie to their lambs, which then carry the infection to other flocks before manifesting the sickness's hallmark symptom, an insatiable itch that makes the beasts scrape their wool off against fixed objects. The USDA has tried unsuccessfully for a half century to eradicate scrapie in the U.S.

Detwiler says USDA concerns about the Vermont sheep escalated last spring after an experimental test on several animals culled from Freeman's flock showed signs of spongiform disease. When brain tissues from the same sheep were sent to an outside lab, four of seven samples tested positive for infectious prions. On July 14, then-Secretary of Agriculture Daniel Glickman issued an emergency order to destroy the two flocks--ironically, just two days after the Faillaces' sheep cheese had won favorable mention on the Martha Stewart Living TV show.

But the situation was still ambiguous. Because the methods of testing for prion diseases are primitive and slow, the USDA couldn't tell whether the sheep had mad cow or scrapie, which is usually addressed nowadays by culling sick animals rather than wiping out entire flocks. Definitive tests require injecting brain tissue from sheep into mice and waiting about two years for telltale lesions that identify the prion strain, a delay the USDA contended was unacceptably risky. On the strength of such ambiguities, Freeman and the Faillaces blocked the seizure order for eight months in a bitter war fought in the courts and media. But in February a federal judge ruled in the USDA's favor.

As the sheep were taken away, Detwiler told reporters that the animals would be trucked to a federal lab in Ames, Iowa, and "humanely" killed. To make absolutely sure no infectious agent escaped, the USDA would liquefy the carcasses in a vat of very strong lye. But doubts will resonate for years. Brain samples from the flocks will be tested in mice for definitive evidence of mad cow; sometime in 2003 the question of whether the sheep really posed a serious risk should be laid to rest.

The threat to humans from spongiform diseases, meanwhile, will continue to morph. The outbreak of mad cow in people intensified an old worry that scrapie, the sheep sickness, might also be able to leap the "species barrier" to humans. Scientists had long discounted that possibility. Scrapie has widely infected sheep for centuries, yet there's no evidence that human spongiform disease occurs at a higher rate in countries with lots of sheep. The Vermont owners cited these reassuring statistics in their fight to save their flocks.

But the science on this issue recently took an ominous turn: A British-French study, published in the March 27 issue of Proceedings of the National Academy of Sciences, a U.S. journal, indicates that at least some cases of human spongiform disease may be due to eating, or coming into close contact with, brain or other tissues from scrapie-infected sheep. The discovery raises the possibility that at least one of the 20-plus known types of scrapie prion has a mad cow-like ability to infect people. If so, the risk is probably low--otherwise more sheep-linked deaths would have shown up in humans by now. Still, if such a link is confirmed, get ready for another international panic, this one about mad sheep.

FEEDBACK: dstipp@fortunemail.com