NEW ORLEANS - As the swine flu threatens to become the next pandemic, the biggest questions are whether its transmission from human to human will be sustained and, if so, how virulent it might become. But even if this virus were to peter out soon, there is a strong possibility it would only go underground, quietly continuing to infect some people while becoming better adapted to humans, and then explode around the world.

What happens next is chiefly up to the virus. It is up to us to create a vaccine as quickly as possible.

Influenza viruses are unpredictable because they are able to mutate so rapidly. That capacity enables them to jump easily from species to species. An avian virus responsible for the 1918 pandemic jumped first from birds to humans, then from humans to swine (as well as other animals). Now, and not for the first time, pigs have given a virus back to humans.

Mutability makes even existing, well-known flu viruses unpredictable. After jumping to a new host, influenza can become more or less virulent -- in fact, different offshoots could go in opposite directions -- before a relatively stable new virus emerges.

Influenza pandemics have occurred as far back in history as we can look, but the four we know about in detail happened in 1889, 1918, 1957 and 1968. The mildest of these, the so-called Hong Kong flu in 1968, killed about 35,000 people in the United States and 700,000 worldwide. Ordinary seasonal influenza, in comparison, now kills 36,000 Americans a year, because the population has a higher proportion of elderly people and others with weak immune systems.

The worst influenza pandemic, in 1918, killed 675,000 in the United States. And although no one has a reliable worldwide death toll, the lowest reasonable number is about 35 million, and some scientists believe it killed as many as 100 million -- at a time when the world's population was only a quarter of what it is today. The dead included not only the elderly and infants but also robust young adults.

All four of the well-known pandemics seem to have come in waves. The 1918 virus surfaced by March and set in motion a spring and summer wave that hit some communities and skipped others. This first wave was extremely mild, more so even than ordinary influenza: Of the 10,313 sailors in the British Grand Fleet who became ill, for example, only four died. But autumn brought a second, more lethal wave, which was followed by a less severe third wave in early 1919.

The first wave in 1918 was relatively mild, many experts speculate, because the virus had not fully adapted to humans. There is very good evidence that people who were exposed during the first wave developed immunity -- much as people get protection from a vaccine.

A similar kind of immune-building process is the most likely explanation for why, in 1918, only 2 percent of those who contracted the flu died. Having been exposed to other flu viruses, most people had built up some protection.

In all four instances, the gap between the time the virus was first recognized and a second, more dangerous wave swelled was about six months. It will take a minimum of four months to produce vaccine in any volume, possibly longer, and much longer than that to produce enough vaccine to protect most Americans. The race has begun.

John M. Barry, author of "The Great Influenza," wrote this article for the New York Times.