By Day 3, Tom Geisbert knew the monkeys were going to die.

He could see it in their faces as he entered the monkey room in Suite AA-4, wearing the blue plastic spacesuit that serves as a uniform for the scientists of the U.S. Army Medical Research Institute of Infectious Diseases when they are working with the world's most vicious viruses. Healthy monkeys in the Biosafety Level 4 lab would often react to his suit by jumping, screeching and beating on their cages. But these monkeys glared sullenly at Geisbert with bloodshot eyes and expressionless faces.

It was Oct. 25, 1999. Geisbert had devoted nearly a decade to finding a vaccine for Ebola. Now he was waiting to see whether the vaccine he'd given the monkeys would provide any protection against the Ebola raging inside their bodies.

He pulled on the back of the cage, and a false wall expanded, slowly pressing a monkey against the bars, where Geisbert jabbed it in the thigh with anesthetic. Fifteen minutes later, he pulled the sedated primate from its cage. He drew blood: The white blood cell count was plummeting. The vaccine for a disease more lethal than smallpox was failing.

It would take four more years of such days before Geisbert and his team would make a breakthrough that could save lives by offering protection from not only an epidemic but also a bioterrorism attack.

Each day, Geisbert steeled himself to witness the ravages of the virus. When he checked on the monkeys on Day 4, the biscuit trays in front of some cages were full; the monkeys had stopped eating. By Day 5, he could see rashes on their arms and chests. On Day 7, the monkeys began to die.

By Day 11, as he dissected the last dead animal, Geisbert was resigned to starting all over again.

Ebola has fascinated Geisbert, 43, since he first looked at the tiny particle through an electron microscope, noting its spaghetti-like shape with the characteristic shepherd's crook on the end. Not long after, he co-discovered a strain that had broken out among research monkeys in Reston -- a tale that made its way into Richard Preston's 1994 bestseller "The Hot Zone" and made Geisbert a celebrity in the science world. He had made little progress since, testing four vaccine solutions, all of which had worked in mice but failed in monkeys.

Geisbert was exasperated, and so was his wife, Joan, who had worked in the lab longer than he had. "We share the same frustration and hard work that went into a lot of failed . . . studies: all the hours, all the time," he recalled. "We have always been, I'd say, just short of obsessed with getting the vaccines."

Geisbert knew that every time he and Joan went into the lab, they were risking their lives. Not that they were doing anything more dangerous than anyone else at the institute, a center for biodefense research at Fort Detrick in Frederick and one of only a handful of labs in the United States allowed to work with deadly, incurable diseases. The PhDs who swarmed the halls dealt with these things every day.

"I think the job that compares closest to what we do would be a fighter pilot -- similar phenotype," Geisbert explained in an e-mail. His search for a vaccine has been reconstructed through interviews, correspondence and a paper published in the June issue of a scientific journal.

He and his wife, a veteran lab technician, gathered these virus jocks into a close-knit team: Lisa Hensley, who'd come to the institute fresh out of graduate school, looking to make her mark; Elizabeth Fritz, who switched specialties just to work with viruses; and Katie Daddario, the youngest, who got into science, she joked, because "I like dead things."

Together with Hensley, Geisbert tallied up what he knew: that Ebola had made the genetic leap from an unknown host animal -- maybe an insect or a bat -- into monkeys in Africa. That the virus then jumped to humans in 1976 via contact with infected animals. That the most recent major outbreak, in 1995 in Congo, had killed 81 percent of the 315 people infected. That the victims had suffered gruesome deaths from massive internal hemorrhaging. And that Ebola had an equally nasty brother known as Marburg, which also had no vaccine.

Geisbert realized they had to break the wall of unknowns surrounding the virus: How does Ebola infect a cell? What does it do when it gets there? Why can't the immune system defend against it?

If his team knew what the virus attacked, it would understand how to defend against it.

He pondered what had gone wrong; the vaccines had all worked in guinea pigs and mice. But the virus acts differently in those species than in monkeys and humans -- the researchers just didn't know how, at least not on the cellular level.

The solution, Geisbert recalled, was obvious: They needed to infect more monkeys. Only this time, they would not let Ebola kill the primates. They would do it themselves.

Geisbert and his team moved around the lab carefully, making sure the curly yellow air hoses plugged into their hips did not become entangled with one another. The air running through the hoses roared inside the suits. It was almost loud enough to drown out the hoo-hoo cries of the younger monkeys and the doglike barks of the older ones. One by one, 21 animals were silenced by anesthetic, pulled from their cages and stuck in the thigh with a needle holding one milliliter of Ebola virus stock.

At the end of that first day, Aug. 6, 2000, the researchers sacrificed three monkeys, putting their small bodies, weighing nine to 14 pounds, on the exam table. Geisbert said the researchers communicated mostly through hand gestures -- a scissors motion or a clicking thumb when one of them wanted a pen -- as they drained the monkeys' blood and took samples of organs.

Before they could infect a single monkey, the researchers needed permission from the institute's Lab Animal Care and Use Committee, which enforces strict international standards on the treatment of animals in scientific research. Geisbert eventually convinced the panel that there would be no other way to understand the virus.

The work was stressful and exhausting, with shifts of 12 hours or more in the constrictive labs. The slightest mistake could ruin the research results, meaning a monkey's life -- valued at roughly $5,000 -- had been wasted. A major accident, such as a needle stick or an animal bite, might cost a human life.

They would do this for six days, killing three or four monkeys a day to gain something like a stop-motion series of photographs of how Ebola destroys the body. Frame 1: Healthy monkeys. Frame 2: Virus detected in the spleen and lymph nodes. Frame 3: The lymph nodes are swollen up to three times their original size, and the virus is floating in the bloodstream. Frame 4: Fevers ranging from 103 to 104 degrees, swollen livers and rashes on the arms and groin. Frame 5: Anorexia, dehydration and depression as the victims start to bleed internally. Frame 6: All the animals are dead.

After months of analyzing lab results, Geisbert and Hensley concluded that Ebola conducts a blitz against the immune system, first knocking out its early-warning radar, known as the dendritic cells.

If his team could find a way of protecting those cells, Geisbert thought, they might be able to give the body a fighting chance.

The solution came over a stein of German beer. Geisbert and a scientist from Canada's Public Health Agency, Heinz Feldmann, were sampling the brews in Marburg, Germany -- where the Marburg virus was discovered -- after a long day at a scientific conference in October 2000.

The hot topic of the conference had been a scientific article published a few months earlier. Viruses are covered in tiny spikes that allow them to stick to cells. The article suggested Ebola's spikes were toxic.

Geisbert and Feldmann technically were competitors in this field, but they were close friends. Over beers, they discussed testing the paper's conclusions.

Feldmann had been working with a virus called vesicular stomatitis, known as VSV, which is harmless in almost all humans. The virus was genetically similar in some ways to Ebola. Thanks to advances in genetic engineering, they could create VSV that grew Ebola's spikes. It was like one person having someone else's fingerprints.

If the spikes weren't toxic, the scientists speculated, they could possibly use the same method to create a vaccine. The immune system would mistake the retooled VSV for Ebola and produce defenses that could deal with it.

There was another advantage. Ebola's spikes preferred to stick to those early warning dendritic cells and infect them. A vaccine armed the same way would protect the cells most in need.

Geisbert left Germany fired up, but he'd have to wait three years before testing the idea. By then, the world had changed.

The slide under Geisbert's electron microscope wasn't Ebola. It was anthrax.

It was Oct. 16, 2001 -- barely a month after the Pentagon had burned and the World Trade Center had been reduced to rubble. Now a new wave of terror was gripping the country: letters filled with deadly anthrax spores.

Geisbert was asked to examine the anthrax powder found in a letter mailed to Sen. Thomas A. Daschle. It was the first of many samples he'd analyze over several months while federal investigators sought the source of the anthrax spores. It was possible that the anthrax had come from the institute itself. Geisbert and his colleagues took polygraphs and answered questions.

Meanwhile, Geisbert's binders full of Ebola research gathered dust in his office, amid the stacks of nature magazines on the floor and the wastebasket full of RC Cola cans. By the time he was able to turn his attention back to Ebola, in February 2002, his team had a lot of catch-up work to do: They were still crunching data from the 21-monkey experiment, thinking about a therapy for Ebola and talking about testing a promising vaccine that used another method.

That vaccine, developed by Gary Nabel of the National Institute of Allergy and Infectious Diseases in Bethesda, worked in monkeys. Nabel published his results in 2003, to much public acclaim.

Geisbert's feelings about Nabel's vaccine were complicated. He had helped Nabel test it in monkeys and said he was thrilled when it worked. Still, he was anxious to try out the idea he and Feldmann had discussed in Germany.

So were some other scientists, who were cautious about the new vaccine. Like any new discovery, there was no guarantee it would work when it reached the human trial stage. Also, the vaccine hadn't been tested against Marburg. Geisbert and Feldmann were almost ready to test a pair of vaccines that could protect against both.

It was time again for the monkeys. Twenty-eight days earlier, six monkeys had been given vaccines: Four were given a vaccine that protected against Ebola, and two were inoculated against Marburg. Now, on Aug. 19, 2003, Geisbert was going to see if the Ebola vaccine worked.

They injected the monkeys with a large, lethal dose of Ebola -- similar to what Geisbert might get if he accidentally infected himself with a needle or was bitten by an animal. The two monkeys who had been given the Marburg vaccine were the control animals.

Then they waited.

Day 3 passed, when Ebola would be expected to appear, and the vaccinated monkeys still looked good. The key was the blood samples: Geisbert didn't want to trust his eyes, worried that the vaccine might only delay Ebola's onslaught. But as the control monkeys turned feverish and their white blood cell counts dropped, the four vaccinated monkeys' counts stayed strong.

By Day 5, Geisbert's confidence was growing. When he walked into the monkey room, the square biscuit trays in front of the vaccinated monkeys' cages were empty. They had healthy appetites, while the control monkeys were beginning to crumple.

Still, the researchers did not want to claim victory too soon. "All four monkeys look great this morning," Geisbert wrote in an e-mail to Feldmann and his colleague, Steven Jones, at 6:40 a.m. on Day 9. "We are at the point now, where each day that goes by where the monkeys show no clinical symptoms of disease, is essentially a logarithmic increase in probability for survival. Continue to keep your fingers crossed."

"I have everything crossed to be honest," Jones replied. "I don't think there are words to describe how I would feel if one of the animals died."

The next day, Geisbert drew blood from the four monkeys. The white blood cell counts were normal. There was no sign of the virus; the monkeys' immune systems had stopped the invasion. They were going to live. The scientists had done it -- they had created a vaccine for Ebola. Their results were published in the June issue of Nature Medicine. The vaccine could make it into a human arm in five years.

The article included the results of one more test -- Marburg. The test, begun Aug. 10, 2004, was identical: Six monkeys, four of them protected, this time against Marburg, and the two controls.

The researchers waited expectantly -- for if this vaccine saved the monkeys, they might be able to give the world a one-shot vaccine for both Ebola and Marburg.

Three days turned into six, and six into nine; the monkeys looked healthy, their blood tests good. When Geisbert came in on Day 10 for another round of tests, he knew it had worked. The biscuit trays were empty.

Tom Geisbert, center, and his team -- Joan Geisbert, left, Katie Daddario, Lisa Hensley and Elizabeth Fritz -- worked for years at the U.S. Army Medical Research Institute of Infectious Diseases, trying to develop an Ebola vaccine. In 2003, they succeeded.Ebola particles are covered in spikes that stick to the immune system's early-warning cells.Lisa Hensley, Elizabeth Fritz, Tom Geisbert and Joan Geisbert were on the team at Fort Detrick, a facility that holds some of the world's most deadly incurable diseases. Special uniforms are needed to protect the scientists from the viruses.