Navajo was the first to die. One day Florena Woody, 21, was healthy; the next day she could no longer breathe. Although many grieved, her abrupt death on May 9, 1993, was not alarming. After all, Florena Woody did have a history of asthma.

But five days later her fiance, Navajo runner Merrill Bahe, 20, died too. Bahe began gasping for air during the 55-mile drive across the state from the couple's trailer in Little Water, N.M., to Woody's funeral in Gallup. His horrified sister-in-law stopped at a general store and dialed 911. Bahe paced, agitated, his skin tinged with yellow, his lips and fingernails turning blue. Then he collapsed.

Autopsies showed that both had essentially drowned, their lungs soaked with serum from their own blood.

And no one knew why.

At the Gallup Indian Medical Center, 55 miles from where Florena Woody died, Bruce Tempest, chief of internal medicine, watched helplessly as Merrill Bahe slowly suffocated in the emergency room.

"He was a physically healthy, very fit person who just died, very rapidly," said Tempest. When an emergency medical technician mentioned that Bahe's fiance had died a couple of days earlier, Tempest became more alarmed. He recalled that he had treated three other patients, all young, previously healthy Navajos, who had died of unexplained respiratory distress during the past six months.

Tempest first suspected a pulmonary form of bubonic plague, a possible culprit in New Mexico, where plague -- a flea-borne disease -- crops up from time to time.

But the results of autopsies on Bahe and one of the other young Navajos were negative for plague, as well as for the common causes of pneumonia and the fast-moving strep-A pneumonia that had killed Muppets creator Jim Henson.

Tempest was perplexed.

The involvement of the lungs suggested some germ that could spread through the air, but why were virtually all the victims young and at the peak of their resistance to disease? Airborne infectious agents can't pick and choose their hosts. Why did the mysterious killer claim Bahe and Woody but spare their 5-month-old son?

Why were there no mild cases?

On May 17, Tempest phoned the state health department and said simply, "I think we have a problem out here."

Eleven days later Jeffrey S. Duchin stepped outside his office at the Centers for Disease Control and Prevention in Atlanta and found his boss Jay Butler striding toward him down the hall.

People are dying in New Mexico, Butler said. It may be an airborne infection or a toxin; it could be a major outbreak. We're leaving in the morning. Do you want to go?

Duchin groaned inwardly. A physician who is training in epidemic control as a member of the CDC's elite two- to three-year Epidemic Intelligence Service program, Duchin had just returned home from two weeks in Florida tracing a fatal Legionnaires' disease outbreak among the touring Welsh Men's Choir to the whirlpool in an Orlando hotel.

He had been looking forward to the long Memorial Day weekend with his family. Nonetheless, he agreed to go.

That evening, Duchin and his colleagues were issued equipment Duchin had never before seen in a doctor's bag: gas masks.

"We had to get them fitted. One at a time, they put us in a room and set off a smoke bomb to see if there were any leaks," he said. "I was scared at first . . . the people who died were just like me -- young, healthy, athletic. As I was walking out to the plane, I was thinking I have a kid and I could die in this outbreak."

The inexplicable deaths of Woody and Bahe, quickly followed by more cases, ignited a nationwide alert over the mysterious new illness. Most of the cases had cropped up in the Four Corners region of the Southwest, where New Mexico, Arizona, Colorado and Utah meet. Dubbed by one newspaper the "Navajo flu," the epidemic exacerbated prejudice against Native Americans.

The story of the search for the mysterious cause of the outbreak, which concluded almost precisely a year ago, involved a collaboration between the CDC, New Mexico health officials, the Indian Health Service and the Navajo tribal health service. It is a textbook example of how epidemiological gumshoes go after a myriad of illnesses, ranging from Legionnaires' disease to E. coli food poisoning to AIDS to the latest strep-A outbreak.

Duchin and two EIS colleagues arrived in New Mexico on Saturday, May 29, for a three-day meeting of more than 30 public health officials.

Arrayed on marker boards around the room were the names of more than 50 suspect germs and toxins: Legionella. Mycoplasma. Chlamydia. Rickettsia. Yersinia. Brucella. Ehrlichia. Influenza. Adenovirus. Herpes. Respiratory syncytial virus. Heavy metals. The pesticide paraquat. Phosphene.

"At the time," Duchin recalled, "there was a great degree of anxiety. We had a deadly illness, a large number of possibilities and no clues."

Case by case, the doctors matched the symptoms to the possible causes. Any known environmental toxins such as pesticides seemed unlikely because all the victims had high fevers. People who die of pesticides don't usually have high fevers. But could it be some new environmental menace?

Hemorrhagic fever was more likely but not a perfect fit. Those in the meeting had never heard of such a virus causing blood vessels to flood the lungs with serum.

So it went for three agonizing days.

By May 31 they had narrowed the list to three possibilities:

* An unrecognized type of hemmorhagic fever virus, an organism that causes illness marked by high fever and leaking blood vessels.

* An atypical influenza.

* An unrecognized environmental toxin.

The day after the meeting in New Mexico ended, a very different group of health care experts met in Window Rock, Ariz. This group had already studied similar episodes, knew the secret of this oubreak and had their own theories on how to prevent it. This meeting was of Navajo medicine men.

Among those attending was Ben Muneta, in whom traditional Navajo medicine and science meet. Muneta's grandfather was a famous medicine man named Hatatli Nez. "I guess he was my role model," Muneta said. "He always had an open mind."

Muneta is also a physician, trained at Stanford and in the CDC's Epidemiology Intelligence Service. Now working for the Indian Health Service, during the epidemic he straddled the scientific world of epidemic control and the spiritual traditions of his frightened, bereaved people.

Muneta and Ron Voorhees, New Mexico's deputy commissioner of health, were the only physicians present at the meeting, hastily convened to discuss the outbreak by the president of the Navajo Nation, Peterson Zah.

"There is a great disharmony in the world caused by people straying away from traditional practices," Muneta said the medicine men told him. "When there is disharmony in the world, death follows."

He spoke to the medicine men, outlining the progress of the investigation -- at that time there had been very little -- and asked for their prayers. The meeting ended. But Muneta sensed the medicine men knew more than they were saying, and they were waiting for him to ask for it in the traditional way.

"Usually if I was asking for knowledge, I would sit for several days during a ceremony," Muneta said. But with the case toll growing, he did not have several days. As tactfully as he could, Muneta kept asking questions.

Finally, the medicine men told Muneta point-blank that although his questions were "quite irregular," they would answer because they knew lives were at stake.

Then they dropped a bombshell.

Something like the Four Corners outbreak had happened in 1918 and 1933, the men told Muneta. In those years, as in 1993, unpredictable weather disrupted the harmony of the world. Unusually harsh winters were followed by excessive spring rains. Pinon trees produced an unusually bountiful crop of pine nuts. Mice, who feast on pine nuts, multiplied. Many young Navajo died.

"My God," Muneta remembered telling himself, with a shock of recognition. "They're talking about the same thing we're talking about."

In essence, the medicine men were describing the mouse as a carrier of the mysterious illness.

The mouse has always been an important creature to the Navajo. It figures in the creation story, in which the revered rodent spreads the seeds of life throughout the world.

But the mouse is also dreaded as a killer of young Navajos and as a carrier of disease. According to legend, when a mouse enters a Navajo dwelling, called a hogan, and sees food lying about, he gets angry at the waste or the sloppiness. As punishment, he chooses the "strongest and finest" young Navajos to die.

The legend is apparently the underpinning of traditions still common among the Navajo. For instance, when a mouse runs over clothing, the clothing is burned.

"Almost every Navajo family has that tradition," Muneta said. "Missionaries used to ridicule the Navajo for burning clothing."

He said this "special warning" about the mouse extends to no other animal. For the first time in his life, Muneta said, he was beginning to understand why.

The hunt for the mysterious killer by the federal and state health officials turned independently to mice. In early June, CDC dispatched seven epidemiologists to the Four Corners region to form an on-site task force that eventually grew to more than 15 members. Duchin and the others scoured medical records, prepared lab specimens to be shipped back to CDC for testing, interviewed patients and their families and "control families" that did not get sick, and set up surveillance operations to track the disease.

The medical investigation began to move rapidly, but the disease outpaced it.

In addition to the Navajo patients, an Icelandic woman named Ragna Halldorsdottir Hooks, who was living near the Navajo reservation, died two days after she contracted what the press had begun to call "the mystery illness." Hooks's death was the first clear evidence that the disease was not singling out Navajos. Soon cases were also cropping up in Arizona, Colorado, Nevada and California.

In Atlanta, C.J. Peters, chief of the CDC's special pathogens research branch, presided over a warren of moonsuits, air locks and safety showers, where the world's most lethal germs are stored and studied.

Peters believed that the epidemiologists were on target when they suspected an unknown hemorrhagic fever virus as the source of the outbreak. He had his own pet picks: two classes of rodent-borne agents known as the hantaviruses, named for the Han River in South Korea, and the arenavirus, a speckled virus whose name derives from the Latin word for sand.

"Originally we suspected these families because the victims were rural people you might expect to have had an exposure to rodents," Peters said.

A simple lab test provided another clue: Many of the patients had a high white-blood-cell count, which is "more likely linked to hantavurus infection than any virus I know of," Peters said.

Yet virologists rarely trust their intuition. Peters and his team pulled vials of 25 disease-causing organisms out of laboratory freezers. The researchers then exposed them to antibodies taken from nine patients with the illness.

Antibodies are proteins produced by white blood cells to attack specific disease-causing organisms. Once a virus-specific antibody is produced, it will attack that virus and only that virus, just the way a magnet will be attracted to metal but not to wood.

In the test, the patients' antibodies zeroed in on a panel of three hemorrhagic fever viruses, from a family known as hantaviruses. The antibodies did not react with any other organism. This suggested that the victims of the Four Corners outbreak had been exposed only to a hantavirus -- and that the rodent-borne hantavirus might indeed be related to the mystery illness.

This preliminary finding was of crucial importance. It enabled field workers to focus their inquiry and look for signs of exposure to the hantavirus. They knew from other outbreaks that mice feces contained the virus and when these droppings were disturbed, say during house cleaning, virus-laden particles became airborne and could be inhaled. It also prompted the CDC to collect mice and test them to see if they carried the same virus as the victims of the outbreak.

The discovery was made on June 4. CDC had been working on the outbreak less than a week.

Soon after Peters linked the virus to mice, he called microbiologist Jamie Childs, one of the few U.S. experts in hantaviruses. "Pack your bags," Peters said.

For the CDC "rat trapper," that's not a simple task. Childs said his four-member team brought five trunks containing "everything we need for the safe processing of animals and transporting them." The team also trained health workers from the Indian Health Service and state health department to do their own trapping.

Wearing sophisticated gas masks and surgical garb, the rat trappers visited patients' homes, trapping mice, killing them and processing blood and tissue specimens so that lab scientists in Atlanta could attempt to determine whether the outbreak was caused by a rodent-borne disease.

Some of the Navajo were irritated by the activity, others puzzled.

"The sight of these people in moonsuits walking around their prairie dog holes and their hogans -- it was a curious sight to say the least. The little kids were laughing. It was like something out of a science fiction movie," Muneta said.

Meanwhile in Atlanta, Peters and his team painstakingly extracted hantavirus genes from tissues snipped from two victims of the disease during autopsy. They then used a relatively new technique known as polymerase chain reaction to multiply strands of DNA millions of times, so that they could compare it with DNA made by other hantaviruses.

To make the connection, scientists count the number of tell-tale acids called nucleotides, the building blocks of DNA, which are matched like twins in a format called base pairs. Working around the clock, the researchers found that the virus taken from the victims was a member of the hantavirus family, but it was unique. Peters and his team had found a newly emergent virus and it had 139 base pairs.

The next step was critical -- proving that the viruses taken from the victims had come from the mice collected on the scene. To make the case, scientists had to compare the genetic makeup of the virus found in the victims with that found in the mice, much like using fingerprints to make the identification.

The rat trappers' first specimens arrived in Atlanta by June 12. Within four days, the researchers had teased out the genetic sequence of the virus taken from the mice.

They then counted the nucleotides: 139 base pairs.

"It was exactly the same virus that was found in the initial patient," said Thomas Ksiazek, chief of the diagnostic section of the CDC's special pathogens branch. "That closed the circle."

Thirty-four days after Florena Woody's death, her killer had been identified.

Writer Steve Sternberg lives in Alexandria.