Two chunky pandas, a male and a female, arrived from China at the National Zoo in Washington on Tuesday. If everything goes as planned, they will eventually have cubs.

Exchanges like this have helped turn giant pandas into the face of conservation worldwide.

The panda program was created with the stated goal of saving a beloved endangered species. Zoos would pay up to $1.1 million a year per pair, which would help China preserve the pandas’ habitat. By following carefully crafted breeding recommendations, zoos would help improve the genetic diversity of the species.

And someday, China would release pandas into the wild.

But a New York Times investigation, based on more than 10,000 pages of documents, has found that the Chinese authorities and American zoos have put a rosy sheen on a program that has struggled, and often failed, to meet those objectives. The records, photographs and videos — many of them from the Smithsonian Institution Archives — offer a detailed, unvarnished history of the program.

They show that, from the beginning, zoos saw panda cubs as a pathway to visitors, prestige and merchandise sales.

On that, they have succeeded.

Today, China has removed more pandas from the wild than it has freed, The Times found. No cubs born in American or European zoos, or their offspring, have ever been released. The number of wild pandas remains a mystery because the Chinese government’s count is widely seen as flawed and politicized.

Along the way, individual pandas have been hurt.

Because pandas are notoriously fickle about mating in captivity, scientists have turned to artificial breeding. That has killed at least one panda, burned the rectum of another and caused vomiting and injuries in others, records show. Some animals were partly awake for painful procedures. Pandas in China have flickered in and out of consciousness as they were anesthetized and inseminated as many as six times in five days, far more often than experts recommend.

Breeding in American zoos has done little to improve genetic diversity, experts say, because China typically sends abroad animals whose genes are already well represented in the population.

Yet American zoos clamor for pandas, and China eagerly provides them. Zoos get attention and attendance. Chinese breeders get cash bonuses for every cub, records show. At the turn of the century, 126 pandas lived in captivity. Today there are more than 700.

Kati Loeffler, a veterinarian, worked at a panda breeding center in Chengdu, China, during the program’s early years. “I remember standing there with the cicadas screaming in the bamboo,” she said. “I realized, ‘Oh my God, my job here is to turn the well-being and conservation of pandas into financial gain.’”

Dr. Loeffler, who spent part of her time in Chengdu as a scholar affiliated with the National Zoo in Washington, said that scientists there used anesthesia excessively and sloppily. At one point, she said, she bucked protocol and jumped onto an examination table to cradle an animal as it was being anesthetized.

Kimberly Terrell, who was director of conservation at the Memphis Zoo until 2017, said, “There was always pressure and the implication that cubs would bring money.” She noted that zoo administrators insisted on inseminating its aging female panda every year, despite concerns among zookeepers that it was unlikely to succeed. It never did.

“The people who actually worked day to day with these animals, who understand them best, were pretty opposed to these procedures,” she said. The zoo said its breeding efforts followed all program requirements. (Dr. Terrell, now a scientist at Tulane University in Louisiana, settled an unrelated gender discrimination lawsuit against the zoo in 2018.)

The Times collected key documents and audiovisual materials from the Smithsonian archives and supplemented them with materials obtained through open-records requests. The trove, which spans four decades, includes medical records, scientists’ field notes and photographs and videos that offer crucial evidence of breeding procedures, side effects and the conditions in which pandas were held.

They show that the riskiest techniques happened in the program’s infancy, but that aggressive breeding continued at the National Zoo and at other institutions for years. A panda in Japan died during sperm collection in 2010. Chinese breeding centers, until recently, separated cubs from their mothers to make the females go back into heat.

Pandas arrived in San Diego this summer, and more will most likely land in San Francisco early next year. There are pandas in a steamy safari park in Indonesia and in an air-conditioned dome in Qatar. So many pandas are in captivity in China that several new tourist attractions are being built.

This panda proliferation has prompted debates among zoo workers and scientists over whether it is ethical to subject animals to intensive breeding when they have no real prospect of being released into the wild. But those discussions have largely played out privately because researchers and zookeepers said that criticizing the program could hurt their ability to work in the field.

Veterinary medicine is always risky, especially with wild animals. When an animal’s life is in danger, the benefits of intervening outweigh the risks. And when a species is on the verge of extinction, conservationists sometimes make a last-ditch effort to save it.

But with pandas, zoo administrators take chances again and again simply to make more cubs, while keeping the grimmest details from the public.

At the center of this story is the National Zoo, which is part of the Smithsonian. Pandas have been part of the zoo’s image since 1972, when President Richard M. Nixon traded a pair of musk oxen for two bears after his historic trip to China.

But the Smithsonian has glossed over the reality of artificial breeding, at times in partnership with the Chinese propaganda apparatus, records show.

American zoos say that keeping and breeding pandas has expanded scientific understanding of the species. “Critical intervention, including conservation breeding, has been necessary for the survival of giant pandas,” the San Diego Zoo said in a statement.

A National Zoo spokeswoman, Annalisa Meyer, acknowledged that efforts to release pandas into the wild were “still developing,” and she said that the program’s success could not be measured in the number of animals released. She said that pandas in zoos were “insurance against extinction” and that animal safety was a top priority.

Western money and attention have also coincided with China’s expansion of nature reserves and stricter logging rules.

Having pandas in zoos also shows that people around the world love, and want to protect, the species, said Melissa Songer, a Smithsonian conservation biologist.

Pandas in captivity are stubborn breeders. Females are fertile for, at best, three days a year. Males can be aggressive or incompetent partners.

But in one of the program’s great ironies, the quest to save pandas may be making it harder for them to breed.

Records show that zoos have long known that keeping pandas in captivity made it less likely that they would mate. Giant pandas in zoos often have a “loss of normal behaviors resulting in reproductive failure,” the National Zoo wrote in an early research proposal.

Heather Bacon, a veterinarian at the University of Central Lancashire, in northwestern England, said humans set the terms. “We choose how they breed. If they don’t want to breed, we make them breed,” said Dr. Bacon, a director of the Bear Care Group, which works closely with zookeepers. “And the justification for that is always, quote-unquote, conservation. Is that a genuine justification?”

“Because all we’re doing,” she added, “is producing more pandas to live in captivity and have those same experiences over and over again.”

The panda program was supposed to fix abuses.

In the 1980s, China sent pandas for short stints to foreign zoos, where they rode bicycles and pushed trollies, like carnival sideshows. Many had been caught in the wild. It took a lawsuit for U.S. regulators to intervene.

After years of negotiation, American zoos and the Chinese government struck a deal, and the U.S. Fish and Wildlife Service issued a policy in 1998. Zoos could rent pandas for a decade at a time, with the money going toward conservation.

American and Chinese scientists also agreed to jointly study panda breeding. The population in captivity showed signs of inbreeding. Artificial insemination efforts had faltered.

So, in the late 1990s and early 2000s, scientists from the National Zoo, San Diego Zoo and other institutions flew to the Sichuan Province of China. Archival photographs and records reveal details of trips that have seldom been discussed but that laid the foundation for breeding around the world.

Researchers shot pandas with tranquilizer darts to anesthetize them, then laid them on stretchers or boards. Bundled up against the cold in spartan concrete rooms, scientists collected semen from the males by inserting electrified probes into their rectums.

They called themselves the “Sperm Team.”

This technique, called electroejaculation, is commonly used in captive breeding. But the scientists drugged some of the animals with unadulterated ketamine, a powerful sedative that veterinarians typically use in combination with other drugs. Ketamine alone can leave an animal anxious and in pain — and partly awake, as a National Zoo veterinarian acknowledged in a presentation at the time.

Some pandas were “light,” meaning they were insufficiently anesthetized, and apparently struggled.

“Animal was light during entire procedure,” JoGayle Howard, a scientist at the National Zoo, wrote in a journal she kept on a 1999 trip. “Almost came off table at one point (used ketamine only this time instead of ketamine and xylazine).”

“Great semen sample with high count,” she added.

During one collection, Dr. Howard wrote that Chinese scientists had quadrupled the voltage to an unsafe 12 volts.

“They used dangerously high voltages and too many stimulations on male Ping Ping after we left,” she wrote. “Male had bloody loose stool and no appetite for months.”

Experts say that electroejaculation should be done cautiously, with minimal voltage. “You can do quite a lot of harm,” said Thomas Hildebrandt, an expert on artificial breeding in animals at Leibniz Institute for Zoo and Wildlife Research in Berlin.

The Chengdu Research Base of Giant Panda Breeding, which today owns one-third of the world’s captive pandas, denied ever using excessive voltage or otherwise harming animals. “We have not had any giant pandas suffer health damage or death during surgery due to the use of ketamine,” the center said in a statement.

Dr. Hildebrandt said that artificial insemination should be done once per cycle, after pinpointing the moment a female is most fertile.

But Chinese scientists inseminated female pandas repeatedly. In one experiment, they inseminated seven females, sedated with only ketamine, as often as six times per animal in five days, meaning the pandas were in and out of stupors.

Notes in the Smithsonian archive show that American scientists accidentally injured one panda’s uterus during an examination. Photographs show pandas vomiting. “Difficult anesthesia,” scientists wrote about a female panda named Lei Lei at a breeding center in Wolong, western China. “Retching and vomiting. Inadequate fasting — food and water. Procedure cut short.”

Many of the scientists from that era have retired or died, and the National Zoo said it had no records of pandas in China being injured. It said that scientists had limited knowledge about panda reproduction at the time. Ms. Meyer, the spokeswoman, said this early research period contributed to improved care and a “panda baby boom.”

Notes make clear that the scientists did not intend to harm the animals. They believed they were saving the species. In conservation efforts, the welfare of the species often trumps that of individual animals.

Dr. Howard became a conservation hero, now memorialized in a Chengdu museum.

But the scientists set in motion a frenzied push to make pandas that continues today.

For decades, the Chinese zoo association has given $1,400 bonuses to breeding centers and zoos for every cub that lives to six months. Those who make “special achievements” can earn up to $7,050.

The Chengdu center’s budget last year included targets for pregnancies and cubs.

That creates an incentive to breed animals as quickly as possible.

In 2017, Lung Yuan Chih, then a researcher with Tsinghua University in Beijing, visited three Chinese breeding centers for her dissertation. All three did multiple electroejaculations or fertilizations on each panda selected for breeding, said Dr. Lung, who is now a director of the Taiwan Human-Animal Studies Institute.

A healthy species has a diverse variety of genes, making it more likely to adapt to illnesses or habitat changes. That is why American scientists helped create detailed recommendations for which pandas should breed.

Those recommendations were often ignored, records show. Instead, the Chinese centers mainly focused on animals that were easy breeders.

Breeding centers also prematurely separated cubs from their mothers.

In the wild, cubs stay with their mothers for 18 months to two years. During that time, females are unlikely to go into estrus, or heat. To make the mothers fertile again, zookeepers have taken cubs away much earlier.

“Sometimes the mothers didn’t have any break at all,” said one Chinese former panda keeper who worked on breeding and spoke on the condition of anonymity because he feared reprisal. “They gave birth every year.”

In the mid 2000s, cubs were moved to nurseries shortly after birth. Later, many were placed with “stepmothers” — essentially panda wet nurses.

Pandas give birth to one or two cubs at a time. Chinese panda enthusiasts who monitor webcam footage documented a female at the Chengdu center in 2017 caring for six cubs.

James Ayala, an American behavioral researcher there, said that the center kept cubs with their mothers whenever possible. Stepmothers are used only when mothers reject their cubs, he said. “Now we know that keeping them with the mom is super, super, super essential,” he said.

Dr. Hildebrandt, the artificial breeding expert, said that he had worked with the center and that practices were improving.

A Times reporter visited Chengdu last month. The center authorized Mr. Ayala to speak but declined to make administrators, scientists or panda keepers available.

During the interview, staff members and local propaganda officials repeatedly interjected to flag topics that were off-limits. Those included the release of pandas into the wild and artificial insemination.

In a recent article titled, “‘Electrocution’ of Giant Pandas! Can It Be True?” the zoo says that artificial breeding is harmless.

When they are old enough, pairs of Chinese pandas are eligible to be rented.

Under the policy governing the rental program, zoos may not profit from pandas.

But records show that, even as the program details were being hashed out, money was at the center of the discussion.

In 1993, zoo representatives from the United States and Europe gathered at the National Zoo to strategize.

The notes from that meeting are full of typos, but they show that zoo administrators were not interested in only displaying a rare species. They wanted cubs, referring to the agreements as “breeding loans.”

“Old males,” said a National Zoo scientist at the meeting, are not “going to bring in as much money as a breeding pair.”

Some attendees acknowledged that shipping pandas around the world did little to protect them. “If we were truly interested in the conservaitonof of the panda,” the notes read, “then we would contribute to them insitu [in the wild] and nont take them out.”

Today, American zoos must submit audits of their panda-related revenue to the Fish and Wildlife Service to prove that they are not profiting. Pandas are expensive. Beyond rent to China, zoos also have to build sophisticated enclosures and buy tons of bamboo.

But pandas attract big donors.

In 1999, before its last pandas arrived, the National Zoo launched a $13 million fund-raising campaign, which included $10.5 million for what it called an “education center.”

An internal document from that period advised employees to deflect a journalist’s questions about the project’s planned gift shop, restaurant, special events area and fund-raising offices. The building is the zoo’s “investment in the future of wildlife on Earth,” the document reads. “So that’s why we want to build the ed facility!”

The zoo, a nonprofit, does not charge for admission. But documents show that it saw pandas as a way to “form strong collaborations with area businesses.”

It brokered panda sponsorship deals with Fujifilm and Animal Planet; worked with local hotels to create packages that included zoo donations; and sourced panda mouse pads, golf balls and shot glasses for the gift shops.

Within months of the pandas Mei Xiang and Tian Tian arriving, one million visitors had come through the gates.

But the pandas struggled.

Scientists have consistently observed panda “stereotypies,” or behaviors associated with captivity. San Diego Zoo scientists studied 47 captive pandas around the world and, in documents submitted to regulators, said that nearly two-thirds did things like “pacing, head tossing, pirouetting and stereotypic cage climbing.”

Conditions in China during those early years may have made things worse. A San Diego scientist wrote to a National Zoo panda keeper that pandas often had problems arising from what he called their “jail cell” stint in “clearly substandard housing.”

For Mei Xiang and Tian Tian, the weather was a challenge. Pandas prefer a cool mountain climate, and by April 2001, the pair languished in the Washington heat.

“Panting,” clinical notes read again and again. The zoo resorted to ice blocks, hosing and air-conditioning. A spokeswoman said that the zoo follows temperature and weather guidelines.

Mei Xiang had irregular stools after being overfed during behind-the-scenes tours, a keeper wrote. When the zoo threw her a party to celebrate her millionth visitor, she slept through it.

As mates, Mei Xiang and Tian Tian were not a great match.

“Tian Tian violently attacked Mei Xiang,” a veterinarian wrote in 2002, after an early mating encounter. Later mating attempts failed.

So staff intervened. Mei Xiang gave birth in 2005 after a single round of artificial insemination.

Subsequent conceptions proved elusive. Scientists began packing multiple procedures into Mei Xiang’s brief fertile window.

Under federal policy, zoos cannot breed pandas simply to make cubs. Zoo notes from that period show that staff were repeatedly reminded that breeding was about science, not cubs.

Administrators tracked the efforts.

“Unfortunately, this was the fourth year in a row that Mei Xiang has not been able to conceive,” the director reported to the zoo’s advisory board in 2010.

The following year was particularly difficult. Mei Xiang vomited after her first insemination. When staff anesthetized her for the second, about 24 hours later, the dart did not fully discharge. Mei Xiang was darted four times that day, leading to a rough recovery.

Ms. Meyer, the National Zoo spokeswoman, said that breeding was closely monitored and followed protocol.

In 2011, the zoo announced that if Mei Xiang failed to produce a cub the next year, it might send her back to China.

Mei Xiang ultimately produced four surviving cubs after at least 21 rounds of artificial insemination. Few of the details were made public, and the Smithsonian has refused to release some information about them through an open-records request.

Years later, in 2022, the Smithsonian Channel made a film about her last cub, “The Miracle Panda,” with a company that is part of China’s propaganda apparatus. It presented artificial breeding as quick, effective and minimally invasive.

The zoo spokeswoman said that filmmakers who needed access to China were required to work with certain production companies. The Smithsonian reviewed the film for “scientific accuracy,” she said.

Almost immediately after each birth, money poured in.

“Overall merchandise sales have increased dramatically,” reads a 2006 document from the zoo’s fund-raising partner.

“Funds much zoo operations, research, education programming,” an employee scrawled on a notepad.

Visitor totals shot up and by 2010, records show, nine out of the 10 best-selling items were panda-related.

Experts say that China typically keeps its most genetically valuable animals in the country. At one point, records show, Tian Tian and Mei Xiang had “the lowest rating” as a pair.

The zoo says that their cubs are healthy and genetically important. “They are part of the breeding program” in China, said Pierre Comizzoli, a Smithsonian reproductive expert who led many of the inseminations. “So this is extremely important.”

At one point, though, records show that experts discussed using a private jet to fly sperm from a panda in San Diego that was a “much more appropriate” genetic match.

“Scientifically, these animals are not important to the population,” Mads Frost Bertelsen, the zoological director at the Copenhagen Zoo, said of the pandas sent overseas. His zoo has pandas, but has not used artificial insemination, he said. “The only reason to do it right now would be a financial one. We would get more revenue if we had cubs.”