Bird flu cannot easily infect humans at present
Tests on a father diagnosed with bird flu in China show he probably caught the disease from his dying son.
Scientists are concerned that if the virus evolves to pass easily from human to human millions could be at risk.
A genetic analysis of the Chinese case published in The Lancet found no evidence to suggest the virus had gained that ability.
But an expert has warned that failure to control outbreaks of disease in poultry is fuelling the risk to humans.
Writing in The Lancet, Dr Jeremy Farrar, of Vietnam's Hospital for Tropical Diseases, said: "Whatever the underlying determinants, if we continue to experience widespread, uncontrolled outbreaks of H5N1 in poultry, the appearance of strains well adapted to human beings might just be matter of time."
However, he said a pandemic was by no means a "biological inevitability".
So far 376 cases of human infection with the H5N1 form of bird flu have been recorded in 14 countries since November 2003, mostly in South East Asia.
There have been 238 recorded deaths from the virus, of which a quarter have come in clusters of two or more linked people.
At present H5N1 remains overwhelmingly a disease which affects birds, and the only humans at risk appear to be those with regular close contact with infected animals.
But the researchers, from the Chinese Centre for Disease Control and Prevention, warn that clusters of H5N1 human infections require close scrutiny to determine whether the virus is starting to evolve into a more potent threat.
The Lancet study highlights the case of a 24-year-old man, from Nanjing, in Jiangsu Province, who died from bird flu.
It is thought he passed the disease on to his 52-year-old father, who survived after receiving prompt medical attention, including anti-viral treatment, and plasma cells from somebody who had been vaccinated against the disease.
It is believed the son caught the disease during a visit to a poultry market six days before he became ill.
Medics also tested 91 people who had come into contact with the infected men, but none of those had contracted H5N1.
The researchers say the possibility that the father caught the disease independently cannot be completely ruled out, but think it unlikely.
Other cases of suspected human-to-human transmission have also been between blood relations.
This may be because related individuals share a genetic susceptibility to infection.
Professor Ian Jones, an expert in virology at the University of Reading, said the fact that the father was the only person to be infected was encouraging.
"The virus isolated had no changes to indicate adaptation to human infection," he said.
"There is no indication from this data that we are any nearer a pandemic."
Professor Wendy Barclay, of Imperial College London, agreed there was no evidence to support the idea that H5N1 had acquired mutations that make it more dangerous to humans.
"The experience here reinforces the idea that there are still several barriers the virus must overcome before it acquires human transmissibility."