The Western Hemisphere is in less danger, according to a study published in the journal PloS Pathogens, which analyzed viruses found in migratory birds sampled from 2001 to 2006 in Alberta and along the Jersey Shore. It found that none carried whole viruses from Eurasian bird pathways.

Therefore, the authors argued, it is more likely that any importation of the virus would be in "birds moved legally or illegally by humans."

It may be even more likely that a human will be the first carrier. There was a close call in early December, when six members of one family in northern Pakistan fell ill, probably infected by a brother who had culled sick poultry.

Another brother, who lived on Long Island, went to Pakistan for the funeral and felt sick when he returned home. He turned out not to have H5N1, but it showed how easily the virus could have reached the Western Hemisphere.

Pakistan had its first human cases last year, as did Laos, Myanmar and Nigeria.

Many small mutations have been recorded that seem to make the virus more adaptable to humans and more resistant to known drugs, but no combination of those producing a superstrain has yet emerged.

Ninety percent of cluster cases have been among blood relatives, and Dr. Arnold S. Monto, an avian flu expert at the University of Michigan School of Public Health, said that suggested a genetic susceptibility that has not yet been defined.

It has long been known that the virus has difficulty attaching to receptors in human noses. A team at MIT has refined that, showing that those receptors come in two shapes, cones and umbrellas, and that avian viruses attach more easily to the cones.

Rapid progress has been made in vaccines. The newest, Monto said, need just small amounts of antigen — 4 micrograms an injection instead of 90 micrograms — making them much more practical to produce.

Some scientists argue for vaccinating millions of people as a precaution. One dose, even if it is based on a 3-year-old strain, might protect against death, if not infection. A second, fully protective dose could be made up from whatever strain has gone pandemic.

Right now, said Dr. Klaus Stoehr, who was chief of flu vaccines for the World Health Organization and now does the same for Novartis, it would take manufacturers about one year to produce a billion doses of any vaccine based on a new pandemic strain. But the pandemic would have circled the globe within three months.

"The peak would be over, and, principally, you'd be vaccinating survivors," Stoehr said. Switzerland, he added, has a vaccine stockpile and plans to test it on soldiers, police officers and health care workers before deciding whether to offer it to all Swiss.

Because of the American swine flu debacle of 1976, in which a vaccine made against a pandemic that never emerged harmed more people than the flu did, experts say they think it is unlikely that many Americans would be willing to take such precautions.