The Five Largest Volcanic Eruptions in North America, and What Their Rocks Reveal

Drive the road from Mammoth Hot Springs toward Norris in Yellowstone National Park and, just past the hoodoos, the highway clings to a cliff in a narrow gorge called Golden Gate. The pale, blocky rock in those walls looks like any canyon sandstone. It is not. It is welded tuff: volcanic ash and pumice that landed so hot, more than two million years ago, that the fragments fused back into solid stone. Every meter of that cliff came out of the ground in a single eruptive episode, the Huckleberry Ridge eruption, and the same event laid a dusting of ash across much of the western half of the United States.

Deposits like that one are how geologists size up eruptions nobody saw. This piece ranks the five largest volcanic eruptions in North America of the ice-age period, and it is less interested in body counts than in rock: how much magma each event pulled out of the crust, what kind of plumbing fed it, how the ground collapsed afterward, and what its ash layer now lets scientists date. If you want the human side of a giant eruption, our story on the Year Without a Summer follows Tambora’s 1815 eruption into New England’s frozen fields. Tambora erupted something like 30 to 40 cubic kilometers of magma. Every eruption on this list was at least ten times larger.

How we ranked them: magma volume, not headlines

Volcanologists measure size in two main ways. The first is bulk volume: how much loose ash, pumice and ignimbrite (the deposit left by ground-hugging pyroclastic flows) an eruption left behind. The second is dense-rock equivalent, or DRE, which squeezes out all the air in that frothy material to estimate how much actual magma left the chamber. A cubic kilometer of fluffy fallout ash can shrink to less than half that once it is converted to DRE, so the two numbers are not interchangeable. The Volcanic Explosivity Index (VEI), devised by Chris Newhall and Stephen Self in 1982, turns bulk volume into a logarithmic scale: each step is roughly ten times more material, and VEI 8, the “supereruption” class, means more than 1,000 cubic kilometers of tephra.

We rank by the best published estimate of erupted magma, using DRE where researchers report it, and we note bulk figures alongside. Every number here carries an uncertainty of tens of percent, because ash blows out to sea, erodes, or lies buried under younger rock. The scope is the Quaternary period, roughly the past 2.6 million years, across the whole North American continent from Alaska through Central America, and only explosive eruptions count. Older, even larger eruptions and the slow-motion flood basalts of the Pacific Northwest are covered at the end.

5. Valles caldera, New Mexico: the twin eruptions that showed how calderas rise again

In the Jemez Mountains west of Santa Fe, two eruptions of almost identical size share fifth place. The first, about 1.61 million years ago, laid down the Otowi Member of the Bandelier Tuff and collapsed the Toledo caldera. The second, dated by argon methods to 1.23 million years ago, erupted about 400 cubic kilometers (roughly 96 cubic miles) of rhyolite as the Tshirege Member and carved the Valles caldera, which cut across its older sibling. A 2016 reassessment in the Bulletin of Volcanology put the Otowi eruption at a minimum of 335 cubic kilometers DRE and probably more than 500 originally, so the order of the two is genuinely uncertain. Both rate VEI 7.

Valles matters to geology out of proportion to its rank. The Jemez field sits where the Rio Grande rift, a zone where the continent is slowly pulling apart, crosses the Jemez lineament, a long northeast-trending belt of volcanic fields. After the Tshirege eruption, magma refilled the reservoir and pushed the caldera floor back up into a resurgent dome, and later lavas oozed out around it in the moat. USGS geologists Robert L. Smith and Roy A. Bailey used Valles as a type example when they described “resurgent cauldrons” in 1968. As a Yellowstone Volcano Observatory column puts it, Valles was one of the first places in the world where resurgence was studied and understood, and lessons learned there were applied to calderas worldwide, Yellowstone included.

After resurgence, thick rhyolite lava domes erupted one after another in the caldera’s moat, the activity migrating counterclockwise around the ring. The last explosive eruption, about 74,000 years ago, laid down the El Cajete pumice, and the Banco Bonito lava followed about 68,000 years ago. The system is quiet now, with rare earthquakes and no significant ground deformation, but the USGS describes a partially molten, mostly solid reservoir 5 to 20 kilometers down.

4. Bishop Tuff, Long Valley, California: six days that defined the zoned magma chamber

About 767,000 years ago, at the eastern foot of the Sierra Nevada, the Long Valley system erupted roughly 600 cubic kilometers (144 cubic miles) of high-silica rhyolite magma, according to the USGS. The roof of the chamber dropped 2 to 3 kilometers, leaving an oval caldera 17 by 32 kilometers. About 350 cubic kilometers of tuff ponded and welded inside the collapsing caldera, up to 200 cubic kilometers flowed out across more than 2,200 square kilometers, and ash fell as far east as Nebraska. Counting all the loose deposits, a 2018 study in Geology put the Bishop Tuff at more than 1,400 cubic kilometers. By magma volume it is VEI 7; by bulk, it brushes the supereruption line.

The Bishop Tuff may be the most studied ignimbrite on Earth. In 1979 Wes Hildreth used it to argue that big silicic magma chambers are stratified, with cooler, more evolved, crystal-poor melt on top and hotter, crystal-richer magma below, and that an eruption unzips the chamber from the top down, preserving that layering in the order the deposits were laid. Field work by Hildreth and Colin Wilson, summarized in the Journal of Petrology in 2007, concluded the whole eruption took about six days with no significant pause, drawing on one integrated reservoir, though other teams have since argued for several separate magma bodies. Its ash is also a prized time marker: precise argon dates on Bishop sanidine have been used to anchor the Matuyama–Brunhes boundary, the last time Earth’s magnetic field flipped, which happened shortly before the eruption.

Long Valley is still warm. Its heat drives geothermal power plants at Casa Diablo, a cousin of the engineered systems in our story on Fervo’s hot-dry-rock plant in Utah, and the caldera floor has been rising on and off since 1978. Seismic imaging published in 2018 suggests the crust beneath it still holds more than 1,000 cubic kilometers of melt, but spread through crystal mush at melt fractions of roughly a quarter, far from an eruptible lake.

3. Los Chocoyos, Atitlán, Guatemala: a subduction-zone supereruption that ties the poles together

The only eruption on this list born at a subduction zone sits under one of the world’s most photographed lakes. Lake Atitlán fills an 18-by-12-kilometer caldera in Guatemala’s highlands, part of the Central American volcanic arc that forms where the Cocos plate dives beneath the Caribbean plate. Its climactic Los Chocoyos eruption, mapped by an international team in the Journal of Quaternary Science in 2021, produced about 1,220 cubic kilometers of tephra, around 900 of it as fallout and 320 as ignimbrite, equal to roughly 730 cubic kilometers of magma. That makes it a VEI 8 and the largest known eruption in Central America. Pyroclastic flow deposits 6 to 10 meters thick lie in Chiapas, Mexico, more than 130 kilometers from the vent.

By bulk volume Los Chocoyos is larger than Yellowstone’s Lava Creek eruption; by DRE it is smaller, which is why it sits third here. Change the yardstick and the two swap places.

Los Chocoyos is also one of the most useful ash beds in the Western Hemisphere. Its glass turns up in sediment cores from the eastern equatorial Pacific, the Caribbean and the Gulf of Mexico, a reach that led a 1980 paper to call it a stratigraphic marker “in three ocean basins.” In 2025, a team writing in Communications Earth & Environment found microscopic shards of it in ice cores from both Greenland and Antarctica, alongside a large sulfate spike. That gives the eruption an age of about 79,500 years and creates the first confirmed supereruption tie between the two polar ice sheets, the kind of time marker that ice-core scientists, like the ones in our Camp Century story, depend on. The same records show no millennial-scale cold spell afterward, which undercuts the idea that it tipped the climate for thousands of years.

2. Lava Creek Tuff, Yellowstone: the eruption that made today’s caldera

About 631,000 years ago, Yellowstone produced the eruption that shaped the park visitors see today. The USGS credits the Lava Creek Tuff with about 1,000 cubic kilometers (240 cubic miles) of erupted material covering 7,500 square kilometers, and the collapse left the Yellowstone caldera, a basin 85 by 45 kilometers. It is VEI 8. In his monograph on the Yellowstone Plateau (USGS Professional Paper 729-G), Robert Christiansen showed the tuff has two distinct sub-sheets and the caldera two ring-fracture zones, each later pushed up into a resurgent dome, today’s Sour Creek and Mallard Lake domes.

Its ash bed is a workhorse of Great Plains geology. The USGS describes windblown Lava Creek ash collecting in ponds and hollows in Nebraska and Kansas up to 9 meters deep, and turning up in Pleistocene deposits from California to Iowa and from Canada to Mexico. For decades, all three Yellowstone ashes were lumped together as the “Pearlette ash.” Only when USGS geologists Glen Izett and Ray Wilcox fingerprinted them by glass chemistry and dating, and mapped them in 1982, did it become clear that they were three separate layers from three separate eruptions. That made each one a precise time stamp for the river terraces and fossil beds that contain it.

New field work is complicating the story. Two ignimbrite units near Sour Creek Dome, mapped decades ago as Huckleberry Ridge Tuff, turned out by dating to belong to the Lava Creek eruption. One survives only as cooled, broken boulders inside a later near-vent deposit, which means it had time to cool completely before the next phase. As the Yellowstone Volcano Observatory explains, that implies a break of years to decades, all before the main Lava Creek deposits were laid down.

1. Huckleberry Ridge Tuff, Yellowstone: the largest of the ice age, and not one single blast

The biggest eruption in North America in the past 2.6 million years is the one in the Golden Gate cliffs. About 2.08 million years ago, the Huckleberry Ridge eruption produced roughly 2,450 to 2,500 cubic kilometers (about 590 cubic miles) of material, buried about 15,500 square kilometers in ignimbrite whose remnants reach from Big Sky, Montana, to Idaho Falls, Idaho, and collapsed a caldera roughly 100 by 50 kilometers, among the largest on Earth. Its ash bed reaches across much of the western half of the country. It is VEI 8, and by volume it is about two and a half times Lava Creek and roughly 2,500 times the 1980 eruption of Mount St. Helens.

It was also probably not a single event. Colin Wilson and colleagues, in a detailed 2019 study summarized by the Yellowstone Volcano Observatory, found that the opening fall deposits at Mount Everts stop and start, with wind ripples, rain and hail marks, and evidence that snow fell during the pauses. Then came three great ignimbrite units, A, B and C. Steam explosions along the A–B contact, where water reached deposits still at 700 to 850 °C, suggest a gap of weeks to a few months. The B–C contact shows the hot C flows chilling against a cooled surface while the deposits below were still venting gas, a pause the researchers estimate at years to a few decades, by comparison with the fumaroles of Alaska’s 1912 Novarupta eruption. A human witness, they suggest, would have seen several eruptions, not one.

The magma was split up, too. Glass and crystal chemistry from the earliest fall layers points to at least four separate melt pockets tapped by different vents. After about 50 cubic kilometers had erupted, the reservoir roof lost support and began to founder, and from then on vents around the collapsing ring tapped eight distinct bodies of melt. That picture, of many discrete pockets in a largely crystalline mush rather than one vast lake of liquid, is now the leading model for how supervolcanoes store magma.

Huckleberry Ridge is also the newest entry in a much longer record. As the North American plate has slid southwest over the Yellowstone hotspot, caldera centers have marched northeast across southern Idaho for some 16 million years, leaving the Snake River Plain as their scar. In 2020, a team writing in Geology identified two previously unknown supereruptions on that track, including Grey’s Landing, 8.72 million years ago, with at least 2,800 cubic kilometers, the largest known eruption of the hotspot. Their work suggests the hotspot’s big eruptions have been getting less frequent, which may mean it is waning.

What the five have in common

Put side by side, these eruptions share a surprising amount. All five erupted rhyolite, sticky, silica-rich magma that traps gas until it fails catastrophically. Each left a caldera rather than a mountain, because the ground fell into the space the magma vacated. Several later pushed their floors back up as resurgent domes. Crystal studies at Long Valley and Yellowstone point to magma that sat in the crust for tens of thousands of years before it erupted, often in separate pockets inside a mostly solid mush. And every one left an ash bed that geologists now use to date the sediments, fossils and ice around it.

They also come from three different tectonic settings: a mantle hotspot under Yellowstone, a stretching continent at Valles and Long Valley, and a subducting plate at Atitlán. Very different ways of melting rock can end in remarkably similar caldera-forming eruptions.

Bigger, older and smaller: putting the list in context

Reach further back and North America’s record grows larger still. During the mid-Tertiary “ignimbrite flare-up,” roughly 36 to 18 million years ago, the Great Basin and southern Rockies hosted dozens of caldera eruptions. The Wah Wah Springs Tuff of Utah and Nevada, about 30 million years old, is estimated in Geosphere at about 5,900 cubic kilometers. The Fish Canyon Tuff from Colorado’s La Garita caldera, about 28 million years old, topped 5,000 cubic kilometers. Its sanidine crystals are now a standard used worldwide to calibrate argon-argon dating, so a sliver of that eruption sits behind countless other ages, including several on this list. The Columbia River flood basalts, mostly 17 to 15 million years old, were effusive rather than explosive, but some single lava flows exceeded 2,000 cubic kilometers.

At the other end of the scale are the eruptions most people have heard of. Mount Mazama in Oregon erupted about 50 cubic kilometers of magma roughly 7,700 years ago and collapsed into the caldera that holds Crater Lake. Its ash is the standard Holocene marker across the Pacific Northwest, and the USGS calls the deposits a natural laboratory for studying zoned magma chambers and caldera collapse. Novarupta in Alaska erupted about 15 cubic kilometers of magma in 60 hours in 1912, the largest eruption of the 20th century anywhere. It revealed that magma can drain sideways: most of it was stored beneath Mount Katmai, whose summit collapsed into a caldera about 10 kilometers from the vent. Mount St. Helens in 1980 produced about 1 cubic kilometer.

Could any of them erupt again?

Yellowstone, Long Valley and Valles are all considered capable of future eruptions, and each still has a hot, mostly crystalline reservoir beneath it. But the USGS stresses that the most likely next eruption at Yellowstone or Valles is a lava flow, not a repeat of the eruptions above, and that any eruption would be preceded by clear changes in earthquakes, ground motion and gas that monitoring networks are built to catch. The biggest lesson from these five eruptions is geological rather than apocalyptic: catastrophic caldera eruptions grow from magma that accumulates slowly, over tens of thousands of years, and the rock they leave behind is one of the best clocks geologists have.

Further reading