Start with the word, because the word is part of the problem. When a lot of people die in a dense crowd, the headline almost always says stampede. It conjures a wild, panicking mob trampling its own, which means, quietly, that the dead brought it on themselves. Here is the truth that crowd scientists have spent decades trying to make stick: it is almost never a stampede, it is almost never panic, and the people who died were, in the most literal sense, standing still when it killed them.
01 · The correctionIt is not a stampede, and it is not panic
A stampede is people running and trampling. It happens, but it is rare, and it is not what fills the mortuaries after a concert or a pilgrimage or a football match. What kills is a crush: people packed so tightly they can no longer move, being compressed by the sheer weight of the crowd around them. The distinction matters enormously, because “stampede” is an accusation. It says the crowd was reckless. Almost every time, the crowd was simply too dense, in a space that was never going to let them out, and that is a failure of planning, not of character.
Panic is the other myth, and it is worth killing outright. The crowd scientist Keith Still puts it as plainly as it can be put: people don’t die because they panic, they panic because they are dying. The lethal pressure builds from geometry and density long before anyone is frightened. By the time there is fear, the physics has already won.
02 · The thresholdWhen a crowd stops being people
Here is the strange, almost physical fact underneath all of it. A crowd is not dangerous because of what people decide to do. It becomes dangerous when it crosses a density at which nobody decides anything anymore. At low densities you move freely. Somewhere around five people per square metre you lose the ability to choose where you put your feet, and contact with strangers becomes constant and involuntary. Push past six or seven people per square metre and the crowd stops behaving like a collection of individuals altogether. It starts behaving like a fluid. Pressure applied at one edge travels through the whole body of people as a wave. These numbers are rough, and sources disagree at the margins, but the direction is not in doubt: past a certain packing, you are no longer a person in a crowd, you are a molecule in a liquid.
03 · The crowd quakeShockwaves you can be thrown by
In 2006, during the Hajj, a catastrophic crush at Mina killed hundreds of pilgrims. The physicist Dirk Helbing and his colleagues did something no one had really done before: they took the video footage and measured the flow. What they found, published in Physical Review E, was a sequence of transitions. First the crowd moved smoothly, what physicists call laminar flow. Then, as density climbed, it broke into stop-and-go waves, the whole mass lurching and halting. And finally, at the highest densities, it tipped into something they called turbulence: violent, chaotic shoving in every direction at once, a “crowd quake.” In that state people were shoved several metres without ever taking a step of their own. You do not fall in crowd turbulence because you stumble. You fall because the fluid you are part of throws you.
The people generating the fatal force are almost always at the back, pushing gently, and they have no idea anyone at the front is dying. In a dense crowd you can feel only the body against you, never the wall of pressure your small nudge is helping to build twenty metres away.
04 · The mechanismWhy people die standing up
Now the part that is hardest to picture, because it contradicts the trampling image so completely. The overwhelming cause of death in a crowd crush is compressive asphyxia. The crowd squeezes the chest so hard from the sides, front and back, that the ribcage and the diaphragm simply cannot expand. You can be fully conscious, fully upright, held bolt upright in fact by the bodies pressed against you, and be entirely unable to draw a breath. It is not the feet of others that kill. It is the pressure of others, stealing the one thing a chest has to be able to do.
Once breathing is genuinely prevented, the clock is brutal. Blood oxygen falls, and unconsciousness follows within roughly a minute or two. Death follows within several minutes more if the pressure is not relieved. The timings vary with how complete the compression is, but the window is short enough that, by the time bystanders even understand what they are seeing, it is very often already too late. This is why the entire discipline is built around prevention. There is almost no such thing as a rescue in time.
05 · The collapseHow a hole spreads through a crowd
The specific way a dense crowd turns lethal usually is not a surge forward. It is a collapse. When people are packed tightly enough to be leaning on one another for support, one person losing their footing leaves a sudden gap. The people around them, off balance and under pressure, fall into that gap, which opens more gaps, and the whole thing propagates outward like a tear, folding people down on top of and against each other. Those at the bottom, or those pinned upright in the wreckage, are the ones who cannot breathe. Crowd researchers call it progressive crowd collapse, and it is why a bottleneck, a blocked exit, or two flows of people meeting head on are so deadly: they are the conditions that raise density to the point where a single stumble takes everyone down.
06 · The recordThe disasters that taught us this
We know all of this because of tragedies, and it is worth naming a few honestly. At Hillsborough in 1989, 97 Liverpool football fans were crushed to death in overcrowded standing pens. In the days after, police briefed the press that drunken, unruly fans were to blame. It was false. The official Taylor Report found the primary cause was a failure of police control, that the fans were not to blame, and that most were not even drunk. It took decades of campaigning to fully overturn the lie, which is exactly why crowd scientists are so fierce about language: the false story does real damage, for years, to the grieving.
The pattern repeats. At Itaewon in Seoul on Halloween 2022, 159 people, most of them young, died in a narrow sloping alley less than four metres wide, with nowhere for the pressure to escape; investigators blamed inadequate preparation despite prior warnings. At the Astroworld festival in Houston in 2021, a surge toward the stage killed 10 people aged 9 to 27, and the medical examiner ruled every single death compressive asphyxia. Different countries, different crowds, the same physics, and the same preventable failure to manage density before it ever became a fluid.
07 · The preventionWhat actually saves lives
Here is the hopeful part, and it is genuinely hopeful: these disasters are almost entirely preventable, and not by controlling how people behave. They are prevented by managing density before it ever reaches the danger zone. That means limiting how many people can enter a space, designing entrances and exits so flows never collide, watching how full an area is getting in real time, and treating a blocked exit as the emergency it is. Crowd scientists are careful to call this density management, not crowd “control”, and they are adamant on one more thing: safety messaging must never blame the attendees. The moment you tell people the dead should have behaved better, you have stopped looking at the thing that actually killed them, which was designed in long before they arrived.
08 · The payoffIf you are ever in one
So, practically, if you ever feel a crowd tighten around you. First, respect density itself as the warning: the moment you can no longer choose where you place your feet, that is the signal to get out, before it gets worse, not after. Keep your feet under you at all costs, because going down is the real danger. If the pressure comes on, bring your hands up in front of your chest like a boxer, forearms across your ribs, to hold a small pocket of space so your lungs can still expand; the CDC and crowd-safety experts both advise exactly this. Do not push back against the crowd, and do not fight the flow, you will only spend energy and add to the pressure. Instead, when you feel a lull, move diagonally toward the edge, going with the movement rather than against it. Stay away from walls, fences and barriers, where the pressure concentrates and has nowhere to go. And remember the one true sentence at the heart of all of it: nobody in that crowd is your enemy, and nobody chose this. It is just physics, and physics is the thing we are supposed to plan around, so that no one ever has to stand there and discover they cannot breathe.
Quick questions
What is the difference between a crowd crush and a stampede?
A stampede is people running in a panic and trampling others, which is what films show and what headlines reach for. It is also rare. A crowd crush is almost the opposite: people are packed so tightly they can barely move at all, and the danger is the crushing pressure itself, not anyone's feet. Crowd-safety researchers push back hard on the word "stampede" because it implies the victims caused their own deaths by behaving badly, when the real cause is almost always density and poor management.
How do people actually die in a crowd crush?
Overwhelmingly from compressive asphyxia (also called traumatic asphyxia): the crowd's pressure squeezes the chest so hard from the sides that the ribcage and diaphragm cannot expand, so the person cannot draw breath. They suffocate, very often while still standing upright, held in place by the bodies around them. Trampling and being crushed underfoot are far less common than the news implies.
Is a crowd crush caused by panic?
Almost never, and this is one of the most important corrections in the whole field. As crowd scientist Keith Still puts it, people don't die because they panic, they panic because they are dying. The pressure builds from density and geometry, a bottleneck, a blocked exit, two flows meeting, long before anyone is frightened. Blaming panic is comforting because it locates the fault in the crowd rather than in how the event was run.
At what crowd density does it become dangerous?
Risk climbs steeply as density rises. Below about 3 or 4 people per square metre you can still move fairly freely. Around 5 per square metre you start to lose the ability to choose where you step, and involuntary contact becomes constant. At roughly 6 to 7 per square metre the crowd behaves like a fluid and pressure can pass through it in waves. These are rough thresholds and sources vary, but the direction is not in doubt: the denser the crowd, the less anyone controls their own body.
What is a "crowd quake" or crowd turbulence?
At very high densities, a crowd stops flowing smoothly and starts moving in sudden, involuntary jolts, first "stop-and-go" waves, then chaotic "turbulent" shoving in all directions. Physicist Dirk Helbing and colleagues identified this by analysing video of the 2006 Mina crowd disaster during the Hajj, describing the transition from smooth (laminar) flow to stop-and-go to turbulence. In that turbulent state people are thrown several metres without taking a step, and it is in exactly these conditions that people fall and are crushed.
Why can't the people pushing just stop?
Because most of them don't know anything is wrong. In a dense crowd you can only feel the person against you, not the front row twenty metres away. Someone near the back leaning in, or a new group arriving at an entrance, adds a small force, and that force sums across hundreds of bodies into a crushing wall at the front. Nobody feels themselves killing anyone. That information gap is a large part of why these disasters happen at all.
What is progressive crowd collapse?
It is the usual way a dense crowd turns lethal. When density is high enough that people are supporting each other, one person losing their footing leaves a gap. The people around them, under pressure and off balance, fall into that gap, which opens more gaps, and the collapse propagates outward like a hole spreading through a packed floor. People end up stacked on and against each other, and those at the bottom or pinned upright can't breathe.
How quickly can a crowd crush kill you?
Fast. Once the chest genuinely cannot expand, loss of consciousness follows within roughly a minute or two as blood oxygen falls, and death within several minutes if the pressure isn't relieved. Estimates vary and depend on how complete the compression is, but the window is short enough that by the time anyone realises what is happening, it can already be too late. This is why prevention, not rescue, is the whole game.
What was the Hillsborough disaster and why does it matter here?
At Hillsborough stadium in Sheffield in 1989, 97 Liverpool football fans were fatally crushed against fencing in overcrowded standing pens. In the aftermath, police briefed the press that drunken, unruly fans had caused it. The official Taylor Report rejected this: it found the primary cause was a failure of police control, that the fans were not to blame, and that most were not drunk. Hillsborough is the landmark case for how a crush gets falsely blamed on its victims, and how long it can take to overturn that lie.
What happened at Itaewon in 2022?
On Halloween night in the Itaewon district of Seoul, an enormous crowd packed into a narrow, sloping alley less than four metres wide. As the density built with no room to disperse, a crowd crush killed 159 people, most of them young adults. An official investigation blamed inadequate preparation and crowd management by authorities, despite prior warnings. It is a textbook example of lethal geometry: a bottleneck with nowhere for the pressure to go.
What happened at the Astroworld festival?
During Travis Scott's performance at the Astroworld festival in Houston in November 2021, a crowd surge toward the stage produced a fatal crush that killed 10 people, aged 9 to 27. The Harris County medical examiner ruled the cause of death compressive asphyxia in every case, an unusually clear confirmation of the mechanism: not trampling, not drugs, but chests unable to expand under crowd pressure.
What actually prevents crowd crushes?
Managing density before it ever gets dangerous: limiting how many people can enter a space, designing entrances and exits so flows don't collide, monitoring how full an area is in real time, and never letting an exit be blocked. Crowd scientists stress this is crowd density management, not crowd "control", and that messaging must never blame attendees. The disaster is almost always designed in long before the event starts, and preventable at that stage.
What should I do if I'm caught in a dangerous crowd?
Keep your feet, and if you feel real pressure, raise your hands in front of your chest like a boxer to protect a small space to breathe (the CDC and crowd-safety experts both advise this). Don't push against the crowd or fight the flow; when you feel a lull, move diagonally toward the edge, going with the movement rather than against it. Avoid walls, fences and barriers, where pressure concentrates. And trust density: if you can no longer choose where you step, leave before it gets worse.
Why do reports keep calling it a stampede if that's wrong?
Partly habit, partly because "stampede" makes a cleaner, more dramatic headline, and partly because it is an easier story: a wild crowd is simpler to describe than a failure of planning. But the word carries an accusation. It implies the dead were reckless. Crowd researchers have spent decades trying to retire it precisely because getting the language right is the first step to getting the prevention right.
Take the quiz on this
A quick 4-question check on what you just read. Get them right to earn XP: no points for just scrolling.
Our sources 14 checked
// every claim on this page was checked before it went up