Images of the devastating earthquake that hit Gujarat, India, last January have yet to fade from memory: buildings reduced to rubble, weeping relatives, the occasional dust-and blood-covered survivor miraculously plucked from the wreckage. That magnitude-7.6 quake, with an epicenter near the city of Bhuj, was India's deadliest ever, wiping out more than 20,000 people. Now, a study published last week in the journal Science says Bhuj may be a mere shiver compared with what lies in store for the subcontinent. The Gujarat quake, says the report, distracts attention from the region where the greatest loss of life should be expected: the 2,100-kilometer Himalayan arc, which stretches from Kashmir in the west to Bhutan in the east. This is the site of the greatest continental collision on earth, where the Indian tectonic plate is ramming northward against Eurasia. Its progress is slow--about 2 centimeters per year--but energy has been building up over the centuries. Eventually the rock must fracture, allowing the Indian plate to lurch northward beneath the Himalayas.
This rupture (read: massive earthquake) is overdue. Geologist Roger Bilham of the University of Colorado in Boulder, the study's lead author, says the fault line running along the whole length of the Himalayas is set to crack. But it won't go all at once. Based on a re-examination of the region's seismic history for the past 300 years and on new GPS satellite data, he and his colleagues estimate that it would take at least seven massive earthquakes, each affecting a 200km to 300km strip of the fault line, to permit the entire plate boundary to slip. Each quake may have a magnitude of at least 8, depending on the timing and severity of the region's most recent disturbance.
Not only would these quakes be more powerful than the Bhuj disaster, but they would occur in more densely populated areas. The danger zone encompasses the capitals of Bangladesh, Bhutan, India, Nepal and Pakistan, and several other cities with populations in excess of 1 million. Several dams, including the massive Tehri hydroelectric dam in Uttar Pradesh, haven't been built to withstand earthquakes. If the Tehri dam broke, it could kill 10 million people. In all, 50 million people are at risk. "If one of the large earthquakes occurred in Delhi, with a population of  million," says Bilham, "the fatality count might be unprecedented. It could be 10 percent of the population."
Of course, the threat of massive earthquakes is not new to India. The country first mapped out its seismological risks 13 years ago, placing Bhuj and areas near Delhi in its highest-risk categories. But India's program to build earthquake-resistant housing has lagged behind. National building codes for the region were put in place a decade ago, but the government still hasn't made them mandatory--even after Bhuj. Those few buildings in Bhuj that were designed and built in line with the new codes survived unscathed, even though the epicenter was just 60km away. But architects and builders tend to ignore the codes. "The codes have been in existence for many years," says Vellaichamy Thiruvengadam, head of building engineering at Delhi's School of Planning and Architecture. "But private companies, as opposed to government organizations, often don't follow them."
Nepal is equally unprepared. Many in Katmandu still recall the massive 1934 quake that killed 4,000 people. Since then, the population has skyrocketed. Unfortunately, construction practices are just as poor as those in Gujarat; since the soil in the Katmandu valley is softer, buildings are even more vulnerable to earthquakes.
Generally, experts in India seem unfazed by Bilham's report, which has already been widely circulated. "No one can predict the timing, location or magnitude of the next earthquake," says Virendra Mittal, director of the seismological division of the Indian Meteorological Department. "Our data goes back only a hundred years," he says. "We'd need thousands to make that kind of statististical prediction."
Bilham and his colleagues are now working on those very data. Digging through ancient Tibetan, Urdu and Arabic texts, the scientists have already discovered at least three previously unknown quakes. One hit Kashmir in 1555, another took place in the central Himalayas 50 years earlier and a third, of which they so far have only one unconfirmed account, may have shaken Nepal in 1255. The point of all this? "Once we get a picture of the last 1,000 to 2,000 years of earthquakes, we might then be able to provide a probability forecast for future events," says Bilham. Meaning that scientists would be able to say there is a 50 percent chance that a strong quake would hit a certain region at a specific time. But that's still far in the future. Until then, Himalayans must brace for the worst.