(FORTUNE Magazine) – TALK ABOUT GUARDING THE HENHOUSE. TO REACH ONE OF the secluded chicken farms of Charles River Laboratories, a Wilmington, Mass., biomedical products and services company, you wind along a New England back road until you come to a gate with a cryptic command: THINK BIOSECURITY. A few hundred feet beyond is a clutch of long, white buildings that house chickens--not just any chickens, but thousands of birds that have been immaculately conceived, hatched, and raised. The hens live in ultraclean inner sanctums, breathe purified air, eat sterilized food, and lay eggs in disinfected plastic nests. To enter their domain, you must first shower and then don a biosafety suit with hat and booties. The worry is that you might infect the birds, not vice versa.

The chickens are a key line of defense in a war that each year kills about ten times as many Americans, on average, as 9/11 did. The foe is influenza. Scientists use the hens' eggs, nearly as pristine as autoclaved glassware, as living chambers for growing influenza seed viruses, the seminal stuff of flu vaccines.

At the farm's main office, a closed-circuit TV shows a vast klatch of the milling hens. "They're very calm today," says Russ Larson, a stocky Iowa transplant with prematurely white hair who is the farm's, well, mother hen. Having grown up on a poultry farm, Larson can tell whether the birds are stressed or ill just by watching them on a monitor, like a mom picking up subtle signs that her toddler is about to have a meltdown.

Larson calls the high-tech chicken coops the "Marriotts of the poultry system." The U.S. has precious few facilities that supply eggs to make flu shots, and details about them "have been identified as a security issue," says a spokesman for Sanofi Pasteur, the vaccine-making unit of France's Sanofi-Aventis. The protectiveness isn't surprising. Flu experts have long warned about a potential replay of the Spanish flu pandemic of 1918, which killed an estimated 20 million to 40 million people around the world. In 1997 a possible forerunner of another such virus appeared: An avian flu strain, known as H5N1, was found to infect humans. At this point it's rarely transmitted between people. But it's very deadly, killing some 70% of those infected, vs. an estimated 2.5% of Spanish-flu sufferers in 1918, and 0.1% of those hit by the typical strains of recent years.

Six people died in the initial H5N1 outbreak in Hong Kong, and now the virus has spread widely in birds across Southeast Asia. Since early 2003 more than three dozen H5N1 victims have died in Thailand and Vietnam. Reports of H5N1's ability to spread among cats and possibly to infect pigs suggest that the fast-mutating virus is feeling its way toward a form that might spread like wildfire in mammals, including humans.

If a pandemic starts, we could face a desperate lack of our main defense against flu--vaccines. That risk was brought home last fall when bacterial contamination at a British flu-vaccine plant owned by Chiron, an Emeryville, Calif., biotech, put the facility out of action, instantly halving America's supply of flu shots and leaving Sanofi as the sole major supplier to the U.S. The vaccine shortage has eased, but Sanofi's plant in Swiftwater, Pa., remains the only American-based source of vaccine for flu shots.

To hear some experts tell it, if there were a flu-catastrophe clock like the one for nuclear doomsday, it would now read two minutes to midnight. At a press conference last November, a World Health Organization official spun out a lurid estimate that if the bird-flu virus evolves into a form that spreads easily among people, perhaps 100 million would die worldwide. That dismaying scenario would resemble the 1918 Spanish flu pandemic, which turned the "old man's friend"--pneumonia--into the young person's horror, striking down mainly people under 50. Victims' lungs filled with bloody fluid, and their faces often turned dark blue from a bodywide lack of oxygen, as if their skin were scorched by inner flames. Blood sometimes flowed from their noses, ears, and eyes. Some fell ill in the morning and were dead by nightfall. Awaiting burial, bodies of flu victims had to be stacked like cordwood at some military camps.

Others counter that if history is any indication, the next flu pandemic won't be as horrific as 1918's--there have been three less severe ones since 1890. (See box, "Is the Risk Overblown?") Yet even a relatively mild pandemic today would likely dwarf disasters such as December's tsunami.

Where would we stand if one started soon?

Fortunately, the threat hasn't been lost on policymakers. Flu-related spending by the U.S. Department of Health and Human Services has risen almost sevenfold over the past five years, to a proposed $283 million for the fiscal year ending Sept. 30. Meanwhile, the flu research budget at the National Institute of Allergy and Infectious Diseases has tripled, to a proposed $61 million for fiscal 2005. The NIAID money has accelerated development of an arsenal to fight flu: handheld biochip devices that in minutes can precisely identify a flu strain from a throat swab, drugs to jam flu viruses' molecular machinery, and bioengineered vaccines churned out by cells cultured in vats called bioreactors, eliminating the complex logistics of supplying millions of eggs for the purpose.

The egg issue is bigger than you'd think. Federal health officials have long fretted that a shortage of eggs might impede emergency vaccine production. Suppliers of fertilized eggs for making vaccines are a rare breed, a cross between farmers and microbiologists. A tour of Charles River's facility gives a glimpse of what it takes: The sequestered hens' freshly laid eggs roll onto conveyor belts and are carried through a kind of air lock to a disinfecting system that looks like a miniature car wash, with sprayers and rotating brushes. Post-washing, the eggs are gingerly placed on plastic flats and conveyed to a processing facility resembling a gym-sized walk-in refrigerator. There machines sort them by weight, and technicians wearing gloves and lab coats inspect them one by one for defects. Nearby, thousands of eggs are stacked on eight-foot-high racks inside a room-sized incubator that warms them for several days until the chick embryos inside are precisely at the right development stage for use in making vaccine. To tell when they're ready, specialists laboriously candle each egg, shining a bright light through its translucent shell to examine the developing embryo. The eggs are then shipped to Sanofi and other makers of human and animal vaccines.

This tricky process won't go away anytime soon. In fact, the federally funded updating of our anti-flu arsenal is still years from full deployment. Meanwhile, we'll be counting on existing defenses, such as isolation of infected patients and restrictions on travel to places with serious flu outbreaks. Mounting a fast, effective response if a pandemic starts may sound like mission impossible after last fall's Chiron debacle. Not necessarily. A crash program to immunize Americans against swine flu in 1976--which at the time seemed about to bring on another 1918 pandemic--shows how a terrible threat can focus the national mind. Despite policy stumbles and technical glitches, it took less than seven months to produce, test, and deploy a novel swine-flu vaccine across the U.S. As it happened, the feared pandemic didn't occur--swine flu turned out to be much less virulent than expected--and the program was branded a fiasco after rare neurological side effects were linked to the vaccine. "But technically it was a phenomenal success," says NIAID director Dr. Anthony Fauci. "I think [such speed] could be repeated." Indeed, the rapid response to SARS in 2003 suggests that health authorities are, if anything, more vigilant than in the days of swine flu.

Several nations, including the U.S., China, Britain, and France, have initiated development of H5N1 vaccines. NIAID expects within weeks to begin a clinical trial with one made by Sanofi-Aventis. And last fall the U.S. government awarded a $13 million contract to Sanofi to produce two million doses of the vaccine, an emergency stockpile that is expected to be at least partly ready by year-end. The U.S. has also stockpiled 2.3 million doses of Tamiflu, an antiviral medicine that could help curtail H5N1's spread. Made by Swiss drug company Roche Group, Tamiflu can speed recovery if taken within 48 hours of the first flu symptoms. It also can lower the risk of infection when taken before exposure--health-care workers near an outbreak would probably get the medicine first.

In order to buy time with such defenses, we'd need to spot a virulent new flu strain before it spreads much. The World Health Organization's flu-surveillance network, which includes 112 labs in 83 countries, serves as the planet's influenza radar. Set up in 1952, it's a well-tuned machine. In 1997 the system played a key role in identifying and stopping the spread of the H5N1 strain in Hong Kong. In 2003 it helped arrest another novel bird flu, H7N7, in Europe. (Like H5N1, the H7N7 strain can sicken humans who are in close contact with infected poultry--it is thought to have killed one of 83 infected people.)

But here's the worry: Flu pandemics are most likely to originate in Southeast Asia, where vast rural populations of people live in close contact with poultry, pigs, and other animal reservoirs of flu viruses that can jump to humans. The surveillance system there is spread thin, and local officials are sometimes reluctant to report possible cases of a disease because publicity can cause devastating economic losses and social upheaval. The Asian surveillance problem was highlighted last year by a probable case of human-to-human transmission of H5N1--an 11-year-old Thai girl, after apparently passing the virus to her mother and aunt while she lay dying, was cremated before specimens could be obtained to test for the virus's presence. Says Scott Harper, an influenza expert at the Centers for Disease Control and Prevention in Atlanta: "Whenever we hear of one case of avian flu [in Asia], we worry that it's just the tip of an iceberg."

Flu's ability to be transmitted before symptoms appear adds to the risk of insidious spread. "If a [newly infected] businessman got on a plane in Hong Kong and came here, his virus could be widespread before we turned around," says Richard Webby, a virologist at Memphis's St. Jude Children's Research Hospital, a major node in the global flu network.

Unless, that is, the bug didn't quite have its killer act together. Let's unfold that. Animal flu viruses that have newly jumped to humans pose high pandemic risk because we lack resistance to them. This is largely because our immune systems aren't familiar with the H and N molecules on their surfaces (whence strains get their names), so we aren't protected by antibodies left over from earlier bouts with human-adapted influenza strains--or from flu shots. It takes the body days to weeks after infection with a novel virus to generate new antibodies and other immune defenses. That gives the bug a good chance to merrily multiply in the lungs, sometimes leading to fatal pneumonia.

Novelty can also work against a flu virus: After first jumping to humans, its molecular tool kit, having evolved to help the virus invade and multiply in animal cells, may still be somewhat ill suited for deftly hijacking human cells. Thus the disease may begin its pandemic career with limited virulence. Something like that apparently happened in 1918: A relatively mild herald wave of influenza preceded the deadly pandemic by several months. The herald virus was probably Spanish flu at an early stage of its evolution into a monster. If that pattern recurs, we would have advance warning of a bad bird flu coming at us.

To scientists at places like St. Jude, H5N1 already seems monstrous--they're not waiting for a herald wave before taking action. Two years ago the Memphis team, led by veteran flu researcher Robert Webster, extracted H5N1 virus from the mucus of an infected Asian patient who had died. Then they isolated each of the virus's eight genes and, with a technique called reverse genetics, employed two of them to fashion a nonvirulent H5N1-like virus to use in making a vaccine. Last spring they learned that the dominant H5N1 strain in Asia had mutated. (Influenza viruses are notoriously changeable, hence the need to update flu vaccines every year.) No problem: Within two weeks they had created a new seed virus based on the altered strain. Packed in a dozen thumb-sized vials, the highly concentrated stuff was shipped to Sanofi, where it was injected into fertilized eggs like those supplied by Charles River. After multiplying in the embryos, the virus was extracted, neutralized, and processed into the vaccine that the NIAID is about to test in the clinic. The trial, expected to be completed near the end of the year, will determine whether the vaccine is safe and what dosage is needed to generate am- ple antibodies targeting H5N1. (The trial's hu- man subjects, by the way, won't be exposed to the deadly virus to see if the vaccine works.)

Even if the H5N1 virus changes again, as it doubtless will, this vaccine should afford some immunity--inoculated people exposed to H5N1 may get ill, but not gravely so. And scaling up production of the vaccine to manufacture the two-million-dose stockpile will give Sanofi's crew invaluable experience. "We have a lot of momentum for making the H5N1 vaccine now and could move to full production fairly quickly," says Sanofi executive James Matthews.

So there's a reasonable chance we could initiate mass vaccinations before an unfolding bird-flu pandemic takes a huge toll. But it probably won't be possible to churn out enough vaccine to quickly cover the entire U.S. population, says David Fedson, a former executive at a Merck/Sanofi-Aventis joint venture in Europe. One reason is that it typically takes at least two doses to engender immunity to a novel virus like H5N1. Sanofi's Pennsylvania plant is geared to make about 150 million doses of H5N1 vaccine over several months. Assuming two doses per person, that would cover 75 million Americans, or about one in four. Several million more might be covered if MedImmune, a Gaithersburg, Md., biotech that makes a nasal- spray flu vaccine, pitched in to fight H5N1.

Could we get more vaccine overseas? Forget it, says Fedson, now a consultant on vaccine issues. "When we have a pandemic, authorities in each country will want to vaccinate their own populations first," he explains. "Given that imperative, vaccine companies will be nationalized," cutting off cross-border sales. No wonder Charles River's special egg operation has the aura of a homeland-defense asset--it is one.

Most other countries will be in a far worse plight than the U.S. if a pandemic hits sometime soon, notes Fedson. A crash program using the world's entire flu-vaccine-making capacity probably could immunize only about 450 million people against H5N1, or 7% of the world population. Further, just nine countries, mostly in Europe, produce over 95% of the flu vaccine made today. That could make for very ugly politics during a mad scramble for vaccine.

New dose-sparing strategies may help, though. Becton Dickinson, a Franklin Lakes, N.J., medical-technology concern, recently introduced syringes that, by minimizing inner waste space, can get up to 25% more shots from a multidose vial than older ones do. Injecting flu vaccine into the skin instead of muscle (the current practice) may conserve vaccine too--relatively low intradermal doses are needed since the skin is especially rich with immune cells that trigger antibody production. A recent study sponsored by Iomai Corp., a biotech in Gaithersburg, indicated that intradermal flu shots would require only a fifth of the standard dose. Dramatic savings are also promised by adjuvants, immune stimulants sometimes added to vaccines to boost efficacy. A study by British drugmaker GlaxoSmithKline suggests that an adjuvant called alum could extend flu-vaccine supplies by a factor of eight. The NIAID plans to test adjuvants, including alum, to stretch experimental bird-flu vaccines that are now in development, says the agency's Fauci.

In an emergency, strategies such as intradermal vaccine injection could come into play quickly, potentially averting shortages, says Benjamin Schwartz, senior science advisor at the CDC's National Immunization Program. That still leaves some burning issues to be addressed, such as who will first get vaccines and antiviral drugs if a pandemic starts. "But we're certainly far more ready today than we were a couple of years ago" to fight bird flu, says Schwartz.

Let's hope he's right. A lull in bird-flu cases in Southeast Asia ended in December, and since then half-a-dozen people with H5N1 infections have died in Vietnam. Recently Cambodia recorded its first known H5N1 fatality, a 25-year-old woman who sought treatment in a Vietnamese hospital and died there. Her 14-year-old brother died before she did, after showing flu-like symptoms. As in several other cases of possible human-to-human transmission, he was cremated before samples could be taken for testing.

FEEDBACK dstipp@fortunemail.com