Maryland health officials, teamed up with a sort of federal medical CIA, are completing their third week on the trail of a microscopic killer: legionella pneumophilia, an elusive bacterium that in the last month left one 61-year-old woman dead and is suspected in the death of a 44-year-old woman and in the pneumonia-like illness of 63 other Charles County residents.

The local officials, aided by health specialists from the national Centers for Disease Control's Epidemic Intelligence Service, are scouring medical records, retesting blood and sputum specimens and delving into the habits of suspected victims of the disease.

Like detectives taking fingerprints at the scene of a crime, the inspectors have taken air and water samples from the homes of more than 30 victims and from the public places those victims or their relatives say they frequented, including churches, shopping malls and schools. They have scraped shower heads, swabbed water droplets from air conditioning vents and sampled the contents of water heaters. They've examined lawn sprinklers and sampled soil from construction sites.

So far, they have come up with nothing, unable to locate a common route of exposure or even decide conclusively that the 64 cases are actually Legionnaires' disease.

"Despite the skill of the sleuth, frequently you cannot find the sources. Maybe something passed through town, maybe whatever disseminated it was a one-shot deal. Maybe that's going to happen here -- maybe the source won't be identified," said Dr. Patricia Charache, director of microbiological laboratories at Johns Hopkins Hospital in Baltimore.

Legionnaires' disease, which strikes an estimated 25,000 Amercians each year, has baffled health officials before. It was named after 182 people attending an American Legion convention at the Bellevue Stratford Hotel in Philadelphia in 1976 became ill, and 26 of them died.

After a year and a half of investigation, the source of infection was identified as the hotel air conditioning system.

The investigators found that the victims included people staying in the hotel and others who were exposed to the bacteria-laden spray from the air-conditioning system when they walked by the hotel.

In another perplexing case in Atlanta in 1981, health officials found that only golfers were striken with the disease. They later traced the source of bacteria to an air conditioning unit that golfers walked by on their way from the 18th hole to the clubhouse, a route that maintenance workers and groundskeepers did not frequent, according to Charache.

Characterized by a rapidly escalating fever of 102 degrees or more, chills and a dry cough, the disease has a 30 percent mortality rate if untreated by antibiotics. It typically strikes people over 30, twice as many men as women, and more smokers than nonsmokers.

The suspected cases have been scattered throughout Charles County, a mostly rural county of about 85,000, bounded on the west by the Potomac River. Officials said yesterday that the legionella bacteria multiply in bodies of water such as ponds and rivers and can be carried through the air on dust particles.

Dr. Richard Goodman, CDC's assistant director of epidemiology, calls the more than 100 doctors, nurses and scientists that make up the Epidemic Intelligence Service, formed in 1951, "shoe leather epidemiologists."

The sleuthing is aimed at finding a common pattern of activity among victims, according to Dr. Max Eisenberg, director of science and health for the Maryland health department.

The specialists sort through mounds of often puzzling information, Goodman said. They list and chart victims' patterns according to time, place and the kinds of people the victims are.

"When were they exposed, when did they get ill, what were they doing?" Goodman said. "Where do they work, go to school? Did they all eat in the same cafeteria within the community or within a building? Who are these people? Do they share membership in a club? Are they children? Are they elderly? Are they women? Are they all pregnant women?"

"You pose questions, which may raise other questions," Eisenberg said. "You're asking people to chronicle their activities in a very detailed fashion for the 10 days before their illness."