Tests in a laboratory recreation of the environment in the nose found that at 32 degrees Celsius, avian flu viruses lose function and cannot spread.
It is likely that the viruses have adapted to suit the warmer 40 degree environments in the guts of birds.
A mutation would be needed before bird flu became a human problem, they said.
Published in the journal PLoS Pathogens, the study also found that human viruses are affected by the colder temperatures found in the nose but to nowhere near the same extent.
In effect, human viruses are still able to replicate and spread under those conditions, the Imperial College London researchers said.
Both viruses were able to grow well at 37 degrees - human core body temperature and equivalent to the environment in the lungs.
They also created a mutated human flu virus by adding a protein from the surface of an avian influenza virus.
This virus - an example of how a new strain could develop and start a pandemic - was also unsuccessful at 32 degrees.
Mutations
Study leader Professor Wendy Barclay said it suggested that if a new human influenza strain evolved by mixing with an avian influenza virus, it would still need to undergo further mutations before it could be successful in infecting humans.
"Our study gives vital clues about what kinds of changes would be needed in order for them to mutate and infect humans, potentially helping us to identify which viruses could lead to a pandemic."
She added further research could point to warning signs in viruses that are beginning to make the kinds of genetic changes for them to jump into humans.
"Animal viruses that spread well at low temperatures in these cultures could be more likely to cause the next pandemic than those which are restricted."
She said swine flu - which was spreading from person to person, seemingly through upper respiratory tract infection - was probably an example of a virus which had adapted to cope with the cooler temperatures in the nose.
Key protein role
Professor Ian Jones, an expert in virology at the University of Reading, said: "This work confirms the fact that temperature differences in the avian and human sites of influenza infection are key to virus establishment.
"It is certainly part of the explanation of why avian viruses, such as H5N1, fail to transmit readily to humans."
He added that the research also showed that the proteins on the outside of the virus were key to its function at different temperatures.
"This helps the monitoring of avian flu as it indicates which changes to look out for."
