Dam safety depends not just on engineering design but on rigorous inspection, timely maintenance, and effective regulatory oversight; historical failures like the South Fork Dam (1889, 2,200 deaths), Teton Dam (1976, 11 deaths), and Edenville Dam (2020) demonstrate that documented warnings, structural problems, and inadequate flood capacity can lead to catastrophic failures when authorities fail to act on known risks, making systematic monitoring and enforcement essential for protecting downstream communities.
Deep Dive
Prerequisite Knowledge
- No data available.
Where to go next
- No data available.
Deep Dive
Top 10 Most Dangerous U.S. Dams! (Here's Why)
Added:A dam is a promise. It is a promise made to everyone who lives downstream that the water behind it will stay behind it.
That the structure holding back a lake, a reservoir, sometimes an entire river, will continue to do that job every day indefinitely, regardless of how routine the morning happens to be for the people living in its shadow. There are more than 91,000 dams in the United States.
The overwhelming majority of them will never fail, but the ones that do fail reveal something specific about what that promise actually rests on.
Inspections that happened or didn't, repairs that were funded or weren't, and warnings that were heard in time or weren't. When a dam fails, there is often very little time between the moment something goes wrong and the moment the water arrives. A dam failure releases the water all at once, a wall of it, moving at speed down a valley where people have lived for generations under the assumption that the dam upstream was doing its job. These are the 10 most dangerous dams in American history, the ones whose failures have shaped the way this country thinks about the question at the heart of every dam safety program. What happens if this doesn't hold? Hit subscribe if you want more like this. We do this every week.
Let's start with the disaster that, more than a century later, is still the deadliest dam failure in American history. A private resort lake above a steel town that had no idea what was coming. The South Fork Dam, Pennsylvania, May 31st, 1889. The South Fork Dam sat above Johnstown, Pennsylvania, holding back Lake Conemaugh, an artificial lake created for the South Fork Fishing and Hunting Club, an exclusive retreat for some of the wealthiest industrialists in America, including Andrew Carnegie and Henry Clay Frick. [music] The dam had been built decades earlier for canal purposes, then abandoned, then purchased and modified by the club. Modifications that included lowering the dam's height and installing fish screens that restricted its ability to release water during heavy rain. On May 31st, 1889, after a day of extremely heavy rainfall, the South Fork Dam failed completely. 20 million tons of water, the entire lake, emptied through the breach in about 36 minutes. A wall of water and debris, at points 40 feet high, traveled down the valley at speeds reaching 40 miles per hour, picking up trees, houses, and people as it went. [music] Nearly 2,200 people did not survive, the deadliest dam failure in American history, a record that still stands. The dam had been showing signs of trouble for years before it failed. Local residents had expressed concerns about the dam's condition to the club. Those concerns were not acted upon. The club's members were never held legally liable for the disaster. Now, imagine Johnstown on the afternoon of May 31st, 1889.
The rain has been falling all day. 14 miles upstream, a private lake owned by some of the richest men in the country, is rising behind a dam that residents have worried about for years. The dam gives way. 20 million tons of water in 36 minutes. 40 feet high, 40 miles an hour, 14 miles to Johnstown. And if the deadliest dam failure in American history happened because the people who could have fixed the problem chose not to, the next one happened because the people who built the dam discovered, the very first time they filled it, that the ground beneath it could not hold the water at all. The Teton Dam, Idaho, June 5th, 1976.
The Teton Dam was an earthen dam on the Teton River in eastern Idaho, completed in November of 1975 by the Bureau of Reclamation. 305 feet high, 3,100 feet across, [music] designed to control flooding, provide irrigation water, and generate hydropower for the Upper Snake River Valley. In June of 1976, during the very first filling of the reservoir, small seeps appeared at the downstream toe of the dam. By June 5th, the seepage had grown. Bulldozer operators were sent to plug what were still believed to be manageable leaks. By 11:55 that morning, the crest of the dam collapsed and the full breach began. Over 1 million cubic feet of water per second poured through the [music] gap, 80 billion gallons in total, draining the entire reservoir in about 6 hours. The flood wave reached 30 feet high in places and moved at speeds up to 20 miles per hour, as much as 5 miles wide at points. 11 people did not survive. 2/3 of Rexburg's 9,000 residents lost their homes. The total damage was estimated at over $2 billion.
The cause was traced to the geology beneath the dam, jointed, permeable [music] volcanic rock that allowed water to pipe through the embankment in a way the design had not accounted for. The dam failed not because it was old or neglected, but because the first time it was asked to do the job it was built for, the ground underneath it could not hold. Now imagine the Teton Dam on the morning of June 5th, 1976.
The reservoir is filling for the first time. Small leaks have been spotted.
Bulldozers are working to plug them. By noon, the dam is gone. 80 billion gallons, 5 miles wide, 20 miles an [music] hour. Rexburg is 15 miles away.
The water arrives in under an hour. And if a federal dam failed on its first fill because of what was hiding in the rock beneath it, the next one is where the warning signs were photographed for years and the dam still failed, killing 39 people on a college campus.
The Kelly Barnes Dam, Georgia. November 6th, 1977. [music] The Kelly Barnes Dam sat above the campus of Toccoa Falls Bible College in Toccoa, Georgia. An earthen dam, originally built around 1899 as a rock crib structure to power a small hydroelectric plant, then raised over decades using earth fill, eventually reaching about 42 feet. By 1957, the dam was no longer used for power generation.
It continued to hold back a recreational lake directly above a dormitory and a trailer park. In the early morning hours of November 6th, 1977, around 1:30 in the morning, while most of the campus was asleep, the Kelly Barnes Dam failed.
Investigators described a combination of factors: a slope failure on the dam's steep downstream face, internal erosion through an old pipe, and rapid progressive erosion that widened the breach until the dam was gone.
Photographs from years before the failure had already shown a slope failure on the downstream face, documented but not addressed. 39 people did not survive. Many in the trailer park and dormitory, asleep when the water arrived with no warning at all.
The disaster led directly to the formation of the Association of State Dam Safety Officials. Now, imagine the Toccoa Falls campus at 1:30 in the morning. The dormitory and trailer park are below the dam, the way they have been for years. The dam has had a visible slope failure on its downstream face. There are photographs of it.
Nobody who is asleep right now knows that. The dam fails. There is no time between the failure and the water arriving. And if a 42-ft earthen dam with a documented, photographed structural problem killed 39 people, the next one is where the documented problem was a federal license formally revoked 2 years before the dam failed. The Edenville Dam, Michigan, May 19th, 2020.
The Edenville Dam was a 96-year-old earthen dam on the Tittabawassee River in central Michigan, one of four dams owned in a series by the same private operator, Boyce Hydro. In 2018, the Federal Energy Regulatory Commission revoked Edenville's license to generate hydroelectric power. FERC determined the dam could only withstand about 50% [music] of the probable maximum flood, the benchmark for the largest flood that location could reasonably face. In May of 2020, days of record rainfall pushed the water level in Edenville's reservoir, Wixom Lake, to a height about 3 ft above the previous historical high.
On May 19th, the embankment failed through static liquefaction. A section of loose sandy soil suddenly lost its strength entirely under the unprecedented water level. The breach released the contents of Wixom Lake down the Tittabawassee River. Two to three hours later, the surge reached the Sanford Dam downstream and overtopped [music] it, causing a second failure.
More than 11,000 residents were evacuated. Remarkably, nobody died.
Officials evacuated roughly 10,000 people about 18 hours before the failure became certain, a decision credited with saving an estimated 10 to 20 lives. Now, imagine central Michigan on May 19th, 2020. [music] 11,000 people have been evacuated from around a dam whose federal license was revoked because it could not handle a flood like this one. The dam fails anyway. Then the next dam downstream fails, too. The warning, two years earlier, had been specific. The failure matched it exactly. And if a dam that failed exactly the way its revoked federal license predicted tells you something about the gap between a warning and a fix, the next one is where the warning came from the dam itself, visibly, publicly, on live television as the spillway of the tallest dam in America came apart in real time.
Oroville Dam, California, February 2017.
Oroville Dam on the Feather River in northern California is the tallest dam in the United States, 770 feet high, completed in the 1960s, holding back a reservoir that supplies water to more than half of California's population. In February of 2017, after heavy rain following a severe drought, water flowing down Oroville's main concrete spillway began eroding a hole that grew rapidly. Engineers shut down the main spillway, which meant the rising reservoir had to be released through the emergency spillway, an earthen hillside never used for a major release in the dam's history. Within about an hour, it, too, began eroding rapidly, cutting a deep gully that threatened to undermine the structure supporting the spillway's concrete weir. On the evening of February 12th, officials ordered the immediate evacuation of nearly 200,000 people, one of the largest dam-related evacuations in American history.
>> [music] >> The emergency spillway held, barely, and water levels were brought down before the erosion reached the point of no return. The subsequent independent investigation found that the crisis was caused by long-term and systemic failures by the dam's owners and regulators to recognize design flaws that had existed since the 1960s.
Repairs cost more than $1.1 billion.
Now, imagine Oroville on the evening of February 12th. The emergency spillway is eroding toward the structure that holds the main spillway in place. Nearly 200,000 people are told to leave immediately. The spillway holds, barely.
The reasons it almost didn't go back 50 years. And if the tallest dam in America came within an hour of catastrophic failure because of flaws there since construction, the next one is where the flaws were never hidden. They are built [clears throat] into the geography itself, in a canyon where the only thing standing between the Colorado River and one of the fastest-growing regions in America is a single arch of concrete poured during the Great Depression.
Hoover Dam, Nevada/Arizona, the structure that cannot be allowed to fail. Hoover Dam holds back Lake Mead, the largest reservoir by volume in the United States, on the Colorado River at the Arizona-Nevada border. Completed in 1936, 726 ft high, it is one of the highest-consequence structures in the entire national dam inventory. Hoover Dam is not on this list because of any documented structural concern. It is rigorously monitored and maintained to the highest standard in the American Southwest. It is on this list because of what a failure, however unlikely, would mean. Lake Mead at full capacity holds enough water to cover Pennsylvania a foot deep. Downstream sit Laughlin, Bullhead City, Needles, Lake Havasu City, and the entire lower Colorado River system that supplies tens of millions of people. The dam was built to standards that accounted for earthquakes, flood events, and the geology of Black Canyon, margins that remain a model for large concrete dam construction. But every dam safety expert who has studied the catastrophic failures of the 20th century has pointed to the same lesson. Confidence built over decades of successful operation is not a guarantee. Now, imagine the scale of Lake Mead held behind Hoover Dam, large enough to cover an entire state a foot deep. The dam has performed exactly as designed for nine decades. The engineers who built it accounted for earthquakes and floods and the geology of Black Canyon. The engineers at Oroville thought their dam was fine for 50 years, too. And if the highest consequence dam in America earns its place through scale alone, the next one earns its place through the opposite combination. A dam whose condition has been formally rated poor for years [music] in a state with over a thousand structurally deficient dams nobody outside the inspection reports ever hears about. Secord and Smallwood Dams, Michigan, the other two. When the Edenville and Sanford Dams failed in May of 2020, they were two of four dams owned by the same operator in series along the Tittabawassee River. Secord Dam sits upstream of Edenville.
Smallwood Dam sits between Secord and Edenville. Both were named as examples of dams assessed to be an immediate threat in subsequent state reviews. The four dam system was, before 2020, exactly the kind of infrastructure that exists across the country without most people who live near it ever thinking about it. Aging, privately owned hydroelectric dams holding back lakes communities had built homes around for generations. People had grown up on Wixom Lake, owned property on it, assumed it would be there. After Edenville and Sanford failed, the remaining dams in the chain became the subject of urgent reassessment. If two dams in a four dam series had failed during one rain event, what did that say about the other two? The Interstate Commission found that the entire system had drifted into a condition where immediate threat applied to multiple structures simultaneously. Now, imagine living on Secord Lake or Smallwood Lake in the days after Edenville and Sanford failed downstream. The dam holding back your lake is the same age, the same design era, regulated by the same underfunded state program. It did not fail in May of 2020. The two below it did. Nobody told you before that week that immediate [music] threat was a category your dam might already be in.
And if a four-dam chain revealed that it hasn't failed yet can apply to multiple structures simultaneously. The next one is where the chain is not four dams on one river, but 162 high hazard dams in a single state. The high hazard inventory, multiple states. What not inspected actually means. The National Inventory of Dams, maintained by the US Army Corps of Engineers, classifies dams by hazard potential. Not by condition, but by what would happen downstream if they failed.
A dam in excellent condition can be high hazard simply because of what's downstream of it. A dam in terrible condition can be classified the same way. Across the states with the highest concentrations of high hazard dams, the data reveals something specific. In some states, a meaningful percentage of high hazard dams have not been inspected at all. And others withhold condition information entirely. Some states share almost everything. Others share only a handful of names out of dams numbering in the dozens or hundreds. This is not a hypothetical concern. The Edenville Dam was inspected and rated poor in 2018, two years before it failed. What was missing was not data. It was the funding, the staffing, and the regulatory urgency to act on what the data already said. Now, imagine every high hazard dam in America that has not been inspected recently, or whose condition rating has not been made public.
Somewhere in that group, statistically, is a dam in the same condition Edenville was in 2018. Nobody knows which one.
That is the point. [music] And if an inventory of high hazard dams with gaps in reporting tells you something about the scale of what is unknown, the number two entry on this list is where everything was known, fully predicted by the dam's own regulator, and it failed anyway, sending a second dam into failure with it.
The Edenville Dam, what 50% of the probable maximum flood actually means.
In 2018, the Federal Energy Regulatory Commission said the Edenville Dam could withstand about 50% of the probable maximum flood for its location, a documented benchmark representing the worst flood that location could reasonably experience. It means the dam, by its own regulator's calculation, could handle roughly half of what it might someday be asked to handle.
[music] Two years later, a flood arrived that pushed the reservoir 3 ft above its previous historical high, well within the range the 50% warning had been about. The dam failed through static liquefaction, a sudden catastrophic loss of soil strength in the embankment, exactly the failure mode that inadequate flood capacity produces under unprecedented water levels. The two years between the warning and the failure were not empty. [music] Regulatory authority transferred from federal to state oversight. The state program that inherited oversight was understaffed and underfunded. Those facts describe the gap between a documented warning and a flood that arrived within the range it described.
Now imagine the Edenville Dam in the two years between 2018 and 2020. The federal warning exists in a file. [music] The people living on the lake have no reason to think about a FERC filing from 2018. May of 2020 arrives. The reservoir rises 3 ft above its historical high.
The dam was rated for about half of this. It does about half of its job, and then it does none of it. And if a documented two-year-old warning that predicted almost exactly the failure that occurred is the second most dangerous outcome on this list, the number one entry is the one where there was no warning anyone could have acted on because the disaster happened on the very first day the structure was ever asked to do its job and the ground itself was the flaw. The Teton Dam and what [music] first fill actually means. Every dam has a moment that does not happen again. The first time the reservoir fills, the first time the structure is asked to hold back the full weight of water it was designed for.
Every assumption made during years of design gets tested for the first time all at once. Most dams pass this test without incident because the engineering accounted for exactly this moment. The Teton Dam was designed by the Bureau of Reclamation, the agency responsible for some of the largest and most successful dam projects in American history including Hoover Dam itself. They were among the most experienced dam builders in the world and during first fill, the moment that reveals whether decades of design assumptions hold up against reality, the ground underneath Teton Dam failed, not the concrete. The volcanic rock in the foundation and abutments, jointed [music] and permeable in ways not fully accounted for, allowed the reservoir to find a path through the rock itself eroding a channel that grew from a trickle to a catastrophic breach in hours. There was no 50-year warning.
There was no photographed slope failure.
There was a brand new dam filling for the first time and a flaw in the bedrock that nobody had identified because the test that would reveal it had never happened until it happened [music] catastrophically. 11 people died. 2/3 of Teton Dam lost their homes. $2 billion in damage. The lesson became the foundation of reclamation's modern dam safety program. Now imagine the morning of June 5th, 1976 one more time. The reservoir is filling for the first time.
Small seeps appear. Bulldozers move to plug them. The seeps are coming through the rock beneath it. Rock that nobody in years of surveys identified as a problem. By noon, [music] 80 billion gallons are moving down the valley at 20 miles an hour. So, there it is. 10 dams, 10 different versions of the same promise, and what happened when it didn't hold? Some failed because the people responsible chose not to act on warnings that had existed for years.
Some failed despite warnings that were specific and arrived right on schedule.
Every dam on this list was built to hold. Every one of them was inspected, designed, or operated by people who believed, correctly, for years or decades that it would. There are more than 91,000 dams in the United States.
Most of them are holding right now. It is just waiting behind every dam that has ever been built. Thanks for watching.
Related Videos
BMW Built A Radial Engine So Good It Made The Spitfire Obsolete Overnight
MachineTitans999
123 views•2026-06-18
UÇAK MOTOLARI ÇALIŞMA PRENSİMİ
PistonTV
428 views•2026-06-17
The Bizarre Design Flaw That Ruined The Convair 990
Jet-Deck
631 views•2026-06-19
Why Are Rocket Nozzles Bell-Shaped? Propulsion | Aerospace engineering | GATE | Viru Sir IITian
conceptlibrary
189 views•2026-06-15
US Navy's Helios laser tech
Striketech0310
6K views•2026-06-18
NEW ENGINEERING DESIGN FOR IAM MARWA APPALOOSA FARM @iammarwa
findingian001
443 views•2026-06-17
The Air Force Built a Jet With Wings Swept the Wrong Way
TheAbsurdArchiveYT
639 views•2026-06-16
China Is Building a Machine the World Can’t Stop
TechAIVision-f6p
192 views•2026-06-15











