ON A HOT APRIL MORNING IN THE PHILIPPINES, Alan Nuñez and five of his friends and relatives and I hiked up a dry creek bed toward Mayon, the country's most active volcano. It's a near-perfect cone, with a steam-plumed crater that towers 8,000 feet above a vast, sloping plain. Mayon has erupted 51 times since the 1600s, killing thousands.

On its lower flanks, we stopped at an area laced with gigantic, sharp-edged boulders. Nuñez, a 31-year-old day laborer, pointed to a nearby slope. Here, at around 1 p.m. on February 2, 1993, 75 people, including Nuñez's uncle and cousin, Apolinaro and Cesar Mirandilla, were cultivating vegetables. Suddenly a cloud exploded from the summit and roiled downward, hugging the ground. Nuñez was watching from a mile off, in his home village of Bonga. The cloud, a grayish, cauliflower-shaped mass, reached the vegetable field in about 60 seconds, with a sound like a low-flying airplane. In his Bicol dialect, Nuñez called it an "uson," or glowing avalanche.

"A pyroclastic flow, though technically we'd call this one a pyroclastic surge," translated Nuñez's godfather, Chris Newhall, a 59-year-old American volcanologist who is perhaps the world's leading expert on the fine art of predicting eruptions and getting people out of the way. Pyroclastic flows are hot, fluid-like mixtures of gases, ash, and boulders that can reach temperatures of up to 1,200 degrees Fahrenheit and travel 100 miles per hour. They usually kill by searing and cementing the lungs or, alternatively, by broiling, dismembering, or vaporizing you.

Nuñez went on, describing what unfolded that day. "Stay with me," Apolinaro said to his son as they saw it approach. "At least they will find us together."

"No, I think I can make it," said Cesar, and he ran.

An hour or so later, Cesar arrived in Bonga straddling a water buffalo. He had been caught by the uson's edge. His skin was charbroiled black, with a consistency like that of "fried chicken," said Nuñez. When Nuñez helped Cesar down, the skin on Cesar's hands came off "like gloves." Cesar lived for three days.

When the ash cooled enough, the villagers went to retrieve the dead. Many were never found. Apolinaro had been hurled some 300 yards downslope and was lying tilted backwards with his upper body sticking above the still-smoking debris. His arms were thrown in front of him, bent at the elbows, hands pushing forward. Apolinaro's face, said Nuñez, was filled with "nasubrahan nín takot."

"An excess of fear," Newhall translated.

Horror stories like these are a big part of what drives Newhall, who has worked around Mayon off and on for more than 40 years. Married to a Filipina, he has a huge extended family nearby—Nuñez is the son of an old friend—and has witnessed many terrifying events here himself. Though he formally retired from his longtime post with the U.S. Geological Survey in 2005, Newhall continues to play a key role on the USGS's Volcano Disaster Assistance Program (VDAP), a geological SWAT team that assesses volcanic crises worldwide. He's also an unpaid consultant to the Philippine Institute of Volcanology and Seismology (PHIVOLCS). In 2005, he moved from Seattle to live year-round at Mayon's foot, thinking he might be of some use here.

Colleagues call Newhall "Saint Christopher" because he seems to care more about people than science. Nearly six feet tall, upright and often somnolent, he looks uncomfortable when anyone talks about his achievements. Off duty, he laughs and jokes easily in English and two Filipino dialects. But when talking volcanoes, he speaks with such slow, unsmiling intensity that some people think he's being unfriendly. Newhall forgets who said it, but he likes the quote that goes something like, "The magnitude of a tragedy may be measured by the degree to which it could have been prevented."

"When it comes to mitigating hazards, Chris is the single most influential volcanologist in the world," says C. Dan Miller, a former head of VDAP. Newhall has an uncanny ability to memorize the unique histories and behaviors of individual volcanoes and has been a pioneer in synthesizing what's known about eruption warning signs. After Washington's Mount St. Helens killed dozens when it blew in 1980, he created a novel risk-calculation system to gauge whether it was safe to reapproach the beast; his principles have since become standard tools for volcano assessment. It was Newhall and colleagues who forecast the 1991 explosion of the Philippines' Mount Pinatubo—the world's largest eruption in almost 100 years—spurring a biblical-scale evacuation and saving up to 20,000 lives.

But Pinatubo was a unique scenario in which almost everything went right. Despite advances in forecasting technology, volcanoes will continue to blow unexpectedly. Even the most sophisticated predictors can fail, and they're so expensive and complex that scientists stationed around the planet keep close tabs on only some 50 obvious suspects. In the United States, a 2005 USGS report identified 13 inadequately monitored "very high threat" volcanoes, including Washington's Mount Rainier, which, when it awakes—not if—could kill tens of thousands with little warning.

"I'd like to say we can always save lives," says Newhall, "but we'd be kidding ourselves if we said we really understood what's going on down there. We cannot yet forecast eruptions as well as we can forecast hurricanes. This is partly because we can't see inside the earth the way we can see in the atmosphere. But I'm optimistic. We're getting better all the time."