H5N1 influenza’s origins stretch back to the 1990s.
a picture of chickens at a poultry farm a picture of chickens at a poultry farm
Chickens at a poultry farm Credit: tianyu wu via Getty
In early 2024, the bird influenza that had been spreading across the globe for nearly three decades did something wholly unexpected: It showed up in dairy cows in the Texas Panhandle.
A dangerous bird flu, in other words, was suddenly circulating in mammals—mammals with which people have ongoing, extensive contact. “Holy cow,” says Thomas Friedrich, a virologist at the University of Wisconsin–Madison. “This is how pandemics start.”
This bird flu, which scientists call highly pathogenic avian influenza, or H5N1, is already at panzootic—animal pandemic—status, killing birds in every continent except for Australia. Around the world, it has also affected diverse mammals including cats, goats, mink, tigers, seals, and dolphins. Thus far, the United States is the only nation with H5N1 in cows; it’s shown up in dairies in at least 17 states.
In all of known history, “This is the largest animal disease outbreak we’ve ever had,” says Maurice Pitesky, a veterinary researcher at the University of California, Davis.
The virus, which emerged nearly three decades ago, is now creating upheaval in the poultry and dairy industries and making economic and political waves due to the fluctuating price of eggs. But there’s more at risk here than grocery-store sticker shock. As it has journeyed around the world on the wings of migrating birds, the virus has infected more than 960 people since 2003, killing roughly half of them. Since the start of 2024, it’s infected dozens of people in the United States—mainly farm workers—and it killed its first person stateside in January of 2025.
So far, H5N1 flu hasn’t acquired the key trick of passing with ease from person to person, which is what could enable a human pandemic. For now, both the US Centers for Disease Control and Prevention and the World Health Organization rate the public health risk as low. But the situation could change.
“The thing about this virus is, every time we think we know what’s going to happen, it does something totally unexpected,” says Michelle Wille, a virus ecologist at the WHO Collaborating Centre for Reference and Research on Influenza in Melbourne, Australia. “And that’s the only consistent thing I can say about it.”
One and done—or a “nasty bastard”
Biologically, H5N1 isn’t so different from any other influenza A virus — the type that resides mainly in wild birds, as well as bats, and has occasionally jumped into human populations. It contains eight pieces of genetic material encoding 11 known proteins. Two proteins, the “H” and the “N” ones, stud the virus’s exterior. H stands for hemagglutinin: It sticks to a cell’s sugars so the virus can gain entry. N is for neuraminidase: It allows newborn viral particles to exit the cell.
Diagram of an influenza virus
Influenza viruses share the same basic structure, but diversity within their genes and proteins means that they can mutate and morph to infect different hosts and evade immune defenses. Credit: ttsz via Getty
But there’s lots of possible variety. The influenza A virus has at least 19 options for the H protein and 11 for the N protein, most of which are present in the various flu strains infecting wild waterfowl. H5N1 flu has version 5 of the H protein and version 1 of the N protein.
There are also variants for the other genes. If two different flu viruses meet in a cell that they’ve both infected, they can swap genes back and forth, creating new kinds of flu offspring.
Thus, all sorts of influenza A viruses infect the guts of wild waterfowl, usually without harm to the birds. But the viruses can cause trouble if they move into other creatures.
A few decades ago, scientists thought they had a handle on what would happen if some bird influenza A virus spilled over into other species. In domestic poultry, it could turn nasty, but it was generally a “one-and-done” situation, says Bryan Richards, emerging disease coordinator at the US Geological Survey’s National Wildlife Health Center in Madison, Wisconsin. What happened in past instances was that all the farm birds would die, the virus would run out of hosts—the end.
And the leap from birds to humans is not easily made. Scientists had long assumed that to infect people, an avian influenza A virus would have to trade genes with another virus in an intermediate species, like a pig, to adapt to mammalian biology.
So back in 1996, when domestic geese in Guangdong province, China, came down with H5N1, it was hardly cause for worldwide alarm.
But a year later, in Hong Kong, a 3-year-old boy died after suffering high fever and pneumonia. It took experts from around the world three months to identify the virus. At first, no one believed it was H5N1, says Robert Webster, a virologist and emeritus professor at St. Jude Children’s Research Hospital in Memphis, Tennessee, who led one of the teams that made the ID. A virus with an H5 was supposed to be a chicken virus. But this H5N1 infected 18 people and killed six of them.
“This was a nasty bastard,” says Webster.
Webster and other experts descended on Hong Kong, where they protected themselves by inhaling inactivated H5N1 virus obtained from that first case, as Webster recounts in the Annual Review of Virology. They learned that the boy’s family had visited a live bird market, and testing identified more H5N1-infected birds in those markets and on farms. It had apparently arrived in ducks from China.
“What blew everyone’s mind, in 1997, was that humans clearly got infected with the avian virus, skipping the pig step,” says Friedrich.
Hong Kong killed all the poultry. That particular viral lineage was snuffed out.
Baffled by viral curveballs
But its parent, back in mainland China, remained. And the vast viral lineage it spawned would continue to defy scientists’ expectations. “This wasn’t the one-and-done,” says Richards. “The virus keeps throwing curveballs.”
H5N1 spread from farm to farm. It continued to infect people, usually those in very close contact with their domestic birds. Then, in 2005, the virus lobbed another curveball: It spilled back into wild birds, by now in a form altered enough to be deadly to them—killing thousands of bar-headed geese, gulls, and great cormorants in China’s Qinghai Lake Nature Reserve. “That,” says Richards, “set the stage for where we are today.”
More birds, likely both wild and domestic, brought H5N1 into Europe and Africa. Through genetic mixing and matching, H5 hooked up with other partners, like N8, for a time. In late 2014, migratory birds brought H5N8 from Asia to the Pacific Coast of North America, where H5 also hooked up with an N2, and the outbreak spread across several states before fizzling out.
The virus continued to spread in Asia, Europe, and Africa, usually as H5N8, with a bit of H5N6. In 2020, reports of H5-containing virus infections in wild and domestic birds started to rise. A new variant of the H5 gene, called 2.3.4.4b, was first spotted in the Netherlands. Viruses carrying this H5 seem to have a particular ability to cross over and infect mammals, says Friedrich.
By 2021, the 2.3.4.4b variety of H5 was back with a form of N1. “From there, we started seeing this mass spread event,” says Wille. The virus arrived in North America in late 2021, this time to stay.
The panzootic had begun.
Illustration showing evolution of H5 gene
Avian flu’s H5 gene has evolved as it spread throughout the world. Explore the trends for yourself at the site of the pathogen-tracking Nextstrain project. Credit: Nextstrain.org
Imperfect biosecurity
As birds migrate south for the winter, they bring H5N1 to poultry farms. Most infected chickens will die, and the primary defense is culling. In the US, more than 166 million chickens have been culled since 2022, though a lull in cases led egg prices to drop in early March 2025.
To prevent spread, biosecurity has become the key watchword. For poultry farmers, that means a variety of things such as limiting human interaction with flocks, washing hands and boots, and wearing face masks. But the virus just keeps spilling over from wild birds into farmers’ flocks. Part of the problem, Pitesky says, is that poultry farms are often located near water sources, like lagoons and rain ponds, where migrating birds roost overnight, putting wild and domestic animals in close proximity. It’s a gut virus in wild birds, and it spreads easily through their feces.
In February 2025, the US Department of Agriculture announced allocation of up to $1 billion in additional funds to combat highly pathogenic avian flu, including support for biosecurity, financial relief for farmers and vaccine research. Companies have designed bird vaccines against H5-containing highly pathogenic avian influenza for a couple of decades, updating them as the virus evolved. The USDA announced in January 2025 that it would update its stockpile of vaccines for chickens, and Zoetis of Parsippany, New Jersey, recently created an updated version. It’s based on a strain that was circulating in 2022 and has continued to do so, says senior vice president for global biologics research and development Mahesh Kumar, who works in Zoetis’ Kalamazoo, Michigan, facility.
The vaccine is effective at preventing symptoms and death, but does not prevent infection or viral transmission, Kumar says. Zoetis’ past vaccines have been used in a handful of other nations for poultry and one even was used by the US Fish & Wildlife Service to protect California condors in 2023.
In early 2025, the USDA granted Zoetis a conditional license for that new formula, but this preliminary licensure is just a step along the way to use, not permission to market or sell the vaccine widely. In fact, the US has never allowed widespread poultry vaccination for highly pathogenic avian flu, though poultry receive a number of other vaccines.
PIcture of eggs at grocery store with sign explaining prices and availability
As farmers were forced to cull infected flocks, supplies of eggs were limited and prices climbed. Credit: https://www.gettyimages.com/detail/news-photo/sign-lets-shoppers-know-about-possible-egg-shortages-at-a-news-photo/2192208914
There are concerns that vaccination could push the virus to mutate faster. But the big issue blocking vaccination is that doing so could limit international poultry trade, and the US is a major exporter of live poultry. A vaccinated animal could carry the virus without symptoms, and many nations don’t want birds that might be invisibly carrying H5N1.
To get around that problem, Zoetis’ vaccine has a twist. In preparing inactivated virus, the scientists used the N2 neuraminidase, instead of the N1 that H5 has recently buddied up with. That provides a way to check whether birds have antibodies that would indicate they’ve been exposed to the vaccine’s N2, to the real virus’s N1, or to both.
Still, it is uncertain whether the US would ever broadly deploy an avian flu vaccine. Pitesky says that much of the power rests with farmers who raise broiler chickens for meat and export; broilers make up about two-thirds of US poultry sales. If the broiler farmers aren’t on board, he believes it’s unlikely the USDA would promote vaccination. The decision might end up being made at a state-by-state level, depending on regional poultry industries, suggests Rocio Crespo, a veterinary researcher at North Carolina State University in Raleigh.
Kumar says Zoetis could turn stockpiled materials into ready-to-use vaccine in two months or less, depending on how close to finished form the vaccine is in storage. “We want to be ready,” he says.
Spillover to dairies
And now the poultry industry’s catastrophe has become the dairy industry’s problem, too. The virus’s 2024 appearance in Texas dairies was a surprise for flu experts: “The literature suggested that dairy cows don’t get influenza A’s,” says Pitesky — but, “as the joke goes, cows don’t read the literature.” Dairies were caught off guard, without the guidelines and support systems that exist for poultry. And by some reports, they’ve been slow to adopt biosecurity measures.
Cows infected with H5N1 usually survive, though they must be taken out of the regular population and spend weeks in a hospital barn. Inflammation in their udders, or mastitis, turns their milk thick and yellowish; splashes of contaminated milk in the milking parlors create potential for the virus to move from animal to animal. (One study suggested that more widespread or respiratory infection does not occur, and there’s no sign yet that beef cattle have been affected.)
The USDA now requires the 48 contiguous states to test milk for H5N1. That testing identified two new spillovers of H5N1 into dairy herds, in Nevada and Arizona, reported in February of 2025.
Cows on a milking carousel
Dairy parlors, such as this one featuring a rotating platform, create plenty of opportunity for the flu virus to spread via infected milk. Credit: stockbusters via Getty
And, worryingly, that virus was a different version than the one that infected cows in 2024. That 2024 spillover featured an H5N1 with a particular collection of flu gene sequences, still H5 2.3.4.4b, called B3.13. But flu viruses evolve rapidly, and that H5 2.3.4.4b has shuffled genes with other viruses more than once, creating lots of variants and subvariants. More recently, another variant called D1.1 has been spreading in wild birds. While B3.13 still accounts for most cattle infections, it’s D1.1 that hopped into dairies in early 2025.
The long-term implications for cattle of D1.1, and avian flu in general, aren’t yet clear. “We’re really hoping this has just been a unique set of circumstances and that we don’t get any more spillover events,” says Jamie Jonker, chief science officer of the National Milk Producers Federation in Arlington, Virginia. But, he adds, “we’d like a vaccine to be in the toolbox and to understand how it can be used.” Zoetis and other companies are working on H5N1 vaccines for cows, though it’s too soon to know if and how such vaccines would be deployed.
Even with vaccines, though, “we may not be able to put out this fire,” says Gregory Gray, an infectious disease epidemiologist at the University of Texas Medical Branch at Galveston. “It appears, to many of us, that these viruses are going to be endemic, or we say ‘enzootic,’ for a long time.”
People on the safe side—for now
What kind of risk does all this pose for people? Gray has studied a number of viruses in cattle and other animals, and he says that while spillovers from one species to another are common, it’s rare that a virus adapts to spread easily in the new species. As of spring 2025, there are no confirmed cases of human-to-human H5N1 transmission in the United States.
“It’s not like in the movies,” Gray says. “It’s going to take continual spillover events for it to get a foothold.” But it can happen, as it did in 2009, when an H1N1 influenza A virus with a novel mix of genes jumped from pigs into people, where it then spread widely. This caused a pandemic, killing an estimated 123,000 to 203,000 people worldwide, a death toll grossly eclipsed by the more than 7 million who died of Covid-19.
To become adept at infecting humans, the virus would have to change the structure of its hemagglutinin. Its current version sticks to a specific arrangement of sugars on the surface of bird cells. The birdlike sugar arrangement is found in cow udders, explaining the mastitis.
Humans do have this birdlike sugar arrangement, but it’s buried deep in the lungs, making the virus hard to catch and hard to spread to another person. It’s also present in human eyes, which could explain why pinkeye was the most common clinical sign in people who caught bird flu in the US in 2024 (many also experienced fever and respiratory symptoms). But for ongoing person-to-person transmission through coughs, sneezes and sniffles, researchers think H5N1 would have to mutate to recognize a sugar arrangement found in the human upper respiratory tract — the nose, nasal cavity, sinuses, mouth, throat and voice box.
It would also have to make changes to the protein that copies its genes, the viral polymerase. This polymerase would need to switch from working well with bird proteins to working well with human ones. It has done that, to some extent: Some versions of H5N1 have acquired relevant mutations that help it replicate in mammal cells. But as of spring 2025, none of the viruses that have jumped from cows to humans have hemagglutinin mutations that are predicted to support person-to-person transmission, Friedrich says.
H5N1 could either evolve on its own, or trade genes with another human-infecting flu. The latter possibility is particularly concerning at times of high rates of seasonal flu, such as during the 2024-25 winter. The more flu virus floating around, the more chances for two kinds to meet in the same cell in the same animal and exchange genes, to birth something new and potentially dangerous.
Factors beyond the virus itself influence pandemic risk, too. “There are a lot of things that have to align, not only on the virus side, but also on the people side,” says Valerie Le Sage, a virologist at the University of Pittsburgh who cowrote an overview of barriers to flu transmission in the 2023 Annual Review of Virology.
One of them is disease history. From recent experiments with ferrets, which get and transmit the virus similarly to the way people do, Le Sage suspects that people who’ve had flu before — that’s most people over the age of 5 — might have enough immunity to stifle the worst consequences of H5N1 flu. In her experiments, ferrets earlier exposed to the 2009 H1N1 swine flu were protected from the worst symptoms and death when later exposed to H5N1 from Texas cattle. Ferrets that were just given H5N1 flu got sick and died. “I can’t tell you exactly how long this protection lasts, but it is nice to see,” says Le Sage.
Also good news is the observation that the virus isn’t hitting anywhere near the reported 50 percent mortality rate in recent US infections. Such rates are imperfect calculations, Friedrich notes, as they are based on people who were sick enough to get tested for H5N1; people who didn’t get very ill would not be tallied as survivors. That would artificially inflate the death rate, though it’s unclear how much this has affected calculations. Asymptomatic infections may not be uncommon, at least in current US cases: A recent CDC study found that three dairy veterinarians had antibodies to H5N1, indicating they’d been infected, but had never noticed symptoms.
The other gene variants that H5N1 has acquired also seem to be a factor, and here the news may be less good. The earlier B3.13 virus seemed to cause mild infections, says David Hamer, a public health epidemiologist at the Boston University Center on Emerging Infectious Diseases. From 2024 through spring 2025, the CDC had tracked 70 H5N1 cases, of any type, in the US, and most have been mild. The one person who died was over 65 and had underlying health conditions — but he also had the newer D1.1 strain, as did a teen in Canada who became severely ill.
Although it’s not fully clear what D1.1 means for people, it could be bad news, speculates Friedrich. “I have this gut feeling, and colleagues of mine do too, that something about the D1.1 genotype may be more permissive for mutations that adapt the virus to humans,” he says.
Issues of trust, and a matter of time
For the general public, the main advice experts offer is to not consume raw milk or undercooked poultry products. Though no human infections from raw milk or undercooked food have been reported to the CDC as of spring 2025, the virus may have been transmitted via raw poultry products in a small number of cases in Southeast Asia, and it has infected cats that drank unpasteurized milk. Pasteurization kills the virus; so does cooking of eggs, chicken and beef.
The US does have some protections ready, including a stockpile of personal protective equipment, antiviral medication — Tamiflu reportedly works on this virus — and the ingredients for making human vaccines. Those ingredients include virus bits, as well as chemicals that help stimulate the immune system. These are stored in bulk, and could be assembled into ready-to-use vaccine doses within weeks to months.
Although those vaccine materials were designed using versions of H5N1 flu from the early 2000s, a recent study suggests that they create an antibody response to the newer 2.3.4.4b versions that have spread globally since 2020, and include both B3.13 and the newly circulating D1.1. Scientists are also working on updated vaccines that would more closely match the virus circulating now.
Social factors could also influence the detection of, and response to, a potential pandemic. Many farm workers are undocumented immigrants, making many reluctant to be screened or seek medical attention. “The population we should be surveilling the most is the population we’re probably not surveilling at a robust enough level,” says Pitesky.
And Friedrich notes the great paradox of the Covid-19 pandemic: It spawned a society that’s less prepared to manage the next outbreak. “The pandemic eroded public trust in science,” he says. “There has been a backlash against the power of public health agencies to do what they need to do to control an outbreak.”
In early 2025, publication of a CDC report on H5N1 flu spreading from cattle to people was delayed. USDA personnel working on bird flu response were laid off; the department later struggled to reinstate them. And $590 million in funding for an RNA-based vaccine (of the kind that proved successful during the Covid pandemic) was put under review. The changes continue, with resignation of a top vaccine official within the US Food and Drug Administration in March and movements starting in April to lay off thousands of federal health workers.
A steep toll on wildlife
Regardless of whether H5N1 jumps from person to person sooner, later or never, it’s raging in wild animals. In the US, thousands of birds of more than 160 native species, including mallards, sparrows, pigeons and bald eagles, have been infected. So have hundreds of mammals of more than two dozen native species, including raccoons, bears and opossums. Some of these get sick, and some die.
Many of these infections are “dead ends,” Richards notes: They don’t pass the virus on. It’s mainly far-flying ducks that have done that.
By late 2022, H5N1 had entered South America and was thundering down the continent’s Pacific coast. “It then traveled the 6,000-kilometer spine of South America in six months, so that’s very fast for a virus that’s not assisted by planes,” says Wille. It hit the tip of South America and jumped to Antarctica.
En route, it killed 40 percent of Peruvian pelicans, at least 24,000 South American sea lions and more than 17,000 southern elephant seal pups.
Wild birds have been affected around the world, and even waterbirds, which normally harbor influenza A without symptoms, have suffered. Though a full census is lacking, individual examples are sobering. The population of great skuas, found primarily in Scotland, is down by a reported 75 percent. An outbreak in California condors in 2023 killed 21 animals, in a species with fewer than 1,000 in existence. “An event like that could change the course of a species,” says Wille. “Are they going to come back or not?”
H5N1 hasn’t reached Australia or New Zealand, but Wille thinks it’s just a matter of time. For the world, the future of this virus, with its propensity to defy expectations, is up in the air. “I think we’re on the precipice of something,” says Wille. “What that something is, I’m not sure.”
This article originally appeared in Knowable Magazine, a nonprofit publication dedicated to making scientific knowledge accessible to all. Sign up for Knowable Magazine’s newsletter.
Knowable Magazine, 2025. DOI: 10.1146/knowable-040125-1. (About DOIs)