A global flu pandemic could kill 62 million people, experts have warned.
Flu viruses are constantly evolving
The 1918 pandemic claimed 50 million lives, and experts in The Lancet predict the toll today would be higher than this, despite medical advances.
The world's poorest nations would be hardest hit, fuelled by factors such as HIV and malaria infections, the Harvard University researchers believe.
Yet developing countries can least afford to prepare for a pandemic, which needs to be addressed, they say.
Lethal global flu epidemics tend to occur three or four times a century.
Some scientists believe a new one may be imminent and could be triggered by bird flu.
So far there have been only 258 cases of the latest strain of avian flu, H5N1, recorded in humans.
But the fear is that this strain could mutate and spread quickly and easily between people, triggering a deadly pandemic.
It is estimated between 50,000 and 700,000 people could die in the UK if such an event occurred.
To forecast how big the global death toll might be, Professor Christopher Murray and his team looked back at the events of the 1918-20 Spanish flu pandemic.
When they extrapolated the mortality rates then to the global population of 2004, they estimated 51-81 million people could die from a similarly severe outbreak and gave a median estimate of 62 million.
And 96% of these deaths would occur in the poorest countries, where there is overcrowding and access to medical care is limited, they predict.
Professor Murray warned: "The burden of the next influenza pandemic will be overwhelmingly focused in the developing world.
"Focusing on practical and affordable strategies for low-income countries where the pandemic will have the biggest effect is clearly prudent."
Dr Neil Ferguson, a UK flu expert at Imperial College London, St Mary's, said: "Access to vaccines, antivirals, and antibiotics for the most vulnerable populations is clearly part of the solution."
But he said it might be more pragmatic to focus on public health measures such as school closure, household quarantine and mask-wearing, although more evidence is needed to show if these can delay or reduce the effect of a pandemic.
"The evidence for non-medical interventions will be eagerly awaited," he added.