How big would an asteroid have to be to destroy Earth?

To physically shatter the planet you would need something the size of a small moon or protoplanet — hundreds to over a thousand kilometres across. To end human civilisation takes far less: an object around 1–2 km wide can trigger global effects, and the roughly 10 km Chicxulub asteroid was enough to wipe out the dinosaurs.

How much energy did the Chicxulub impact release?

The Chicxulub impact 66 million years ago released on the order of 10^23 joules — roughly 100 million megatons of TNT, millions of times more than the largest nuclear weapon ever tested.

What happens if an asteroid hits the ocean?

Because most of Earth is water, a random impact most likely lands in the ocean. It flash-vaporises an enormous volume of water and can raise tsunamis, though deep-water impacts tend to produce smaller coastal waves than once feared.

How often do asteroid impacts happen?

Small Chelyabinsk-class airbursts (about 20 m) happen every few decades; Tunguska-class events (50–60 m) every few centuries to a thousand years; 1 km objects roughly every 500,000 years; and Chicxulub-scale impacts only every hundred million years or so.

Is this asteroid impact calculator accurate?

Every figure is computed from the published Earth Impact Effects equations (Collins, Melosh & Marcus, 2005) — the same first-order models scientists use — so it is a solid approximation rather than a full hydrodynamic simulation.

Asteroid Impact Simulator & Calculator

What this tool is actually doing

Most asteroid impact calculators are one of two things. Either they are beautiful and tell you nothing about where the numbers came from, or they are authoritative and look like they were built in 2005 — because they were. This one is trying to be both, and the way it does that is by refusing to give you a number without also giving you the equation that produced it. Click any figure above.

The physics is not ours. It comes almost entirely from a single paper: Collins, Melosh & Marcus (2005), Earth Impact Effects Program, published in Meteoritics & Planetary Science. It is the same model that sits behind Purdue University's impact calculator, and it is the closest thing the field has to a standard. Where we depart from it, we say so on the page.

The single most important question: does it reach the ground?

Everything else follows from this. An asteroid arriving at twenty kilometres per second meets air that, from its point of view, is a wall. The pressure on its leading face is roughly the density of the air multiplied by the square of the speed — and by the time it is forty kilometres up, that pressure is measured in megapascals.

Rock is strong in compression and weak in tension, and a rubble pile — which is what most asteroids are — is barely held together at all. So the object cracks, then flattens, then spreads. The model calls this the pancake, and the moment it has spread to about seven times its original width is the moment it dumps essentially all of its remaining energy at once. That is an airburst.

This is why size is not the whole story. A fifty-metre rubble pile at seventeen kilometres per second never reaches the ground — that was Tunguska, and it still flattened two thousand square kilometres of forest. A fifty-metre iron body at thirteen kilometres per second punches straight through and digs a kilometre-wide hole — that was Meteor Crater. Same diameter. Completely different day.

What each ring means

The contours on the map are not decoration. Each one is a specific overpressure or heat flux, and each has a specific consequence that has been measured — mostly, uncomfortably, at Hiroshima and Nagasaki, and in nuclear test data from the 1950s.

Contour	Overpressure	What that means
Fireball	—	Vaporisation. There is no survival mechanism here.
20 psi	138 kPa	Reinforced concrete fails. Near-total fatality.
5 psi	34 kPa	Ordinary homes collapse. Historically the deadliest band.
1 psi	7 kPa	Every window in the city breaks.
0.7 psi	5 kPa	Forest flattened — calibrated to Tunguska's blowdown.

That last row deserves a note. The standard tables stop at 1 psi, because below that nuclear-era analysts stopped caring. But asteroid airbursts are slower, longer-wavelength events than bomb blasts, and they do real damage well below the tabulated floor. So the two weakest contours in this tool are not taken from a table at all — they are fitted directly to the two airbursts we have good data for: Tunguska in 1908, and Chelyabinsk in 2013.

Why the casualty figure is a range, and a wide one

The population data underneath this tool is better than you might expect: nearly 45,000 settlements at their real coordinates, plus a modelled grid covering the roughly five billion people who do not live in any gazetteer's list of cities. Nationally it reconciles to the UN's 2024 figures.

The lethality is where it gets soft. Consider Chelyabinsk. Roughly 1,500 people were injured and not one was killed — and almost every injury happened because people saw a brilliant flash, walked to the window to look at it, and were standing there two and a half minutes later when the shockwave arrived and turned the glass into shrapnel. Change one behavioural detail and that number moves by an order of magnitude.

So: the same impact at 3 a.m. on a Sunday and 3 p.m. on a Tuesday differ by a factor of about three. With two days' warning and a functioning evacuation, by far more than that. Read the range. Ignore the middle.

The things this model does not do

Terrain. Blast waves are computed over flat, empty ground. A valley wall is worth a lot; the model does not know it exists.
Fire. Firestorms killed more people at Hiroshima than the blast did. Whether an impact starts one depends on the city, the weather and the season, and no model predicts it reliably.
Tsunami, honestly. Ocean impacts produce short-wavelength waves that disperse quickly and often break harmlessly offshore. The classic model over-predicts run-up badly enough that this tool declines to give you a single wave height.
Ejecta and climate in detail. Chicxulub's severity had as much to do with the sulphate-rich rock it happened to hit as with its size. Geology matters and is not modelled.

How likely is any of this?

The recurrence figure on each result is a long-run average drawn from three decades of bolide detections and the lunar cratering record. It is not a countdown. Impacts are a Poisson process: the fact that Tunguska was a century ago tells you precisely nothing about when the next one is due.

What has changed is that we can now see them coming. Surveys have found the large majority of the kilometre-plus objects — the civilisation-ending class — and none of them is on course to hit us. The unfinished work is in the 140-metre range, the city-killers, where most of the population is still undiscovered. And in 2022, DART demonstrated that if we do find one early enough, we can move it.

That is the genuinely optimistic thing about this page. Every other mass extinction mechanism in Earth's history was unavoidable. This is the only one we have the technology to prevent — provided we look.

Sources

Collins, G. S., Melosh, H. J. & Marcus, R. A. (2005). Earth Impact Effects Program. Meteoritics & Planetary Science 40(6), 817–840.
Chyba, C. F., Thomas, P. J. & Zahnle, K. J. (1993). The 1908 Tunguska explosion. Nature 361, 40–44.
Schmidt, R. M. & Housen, K. R. (1987). Some recent advances in the scaling of impact and explosion cratering.
Glasstone, S. & Dolan, P. J. (1977). The Effects of Nuclear Weapons, 3rd ed.
Brown, P. et al. (2002). The flux of small near-Earth objects colliding with the Earth. Nature 420, 294–296.
Population: GeoNames gazetteer (CC BY 4.0); UN World Population Prospects 2024.
Coastlines: Natural Earth (public domain). Basemap: OpenStreetMap contributors, CARTO.

What would happen if an asteroid hit Earth?

It depends almost entirely on size. Take a 140-metre stony asteroid — a "city-killer" far too small to threaten civilisation, but big enough to erase a metropolis — arriving over a major city at about 17 km/s. At second zero there is a flash brighter than the Sun as it slams into the ground and excavates a crater over a kilometre wide. A fireball expands faster than sound; anyone who turns toward the light does so before the blast wave reaches them. For the next two to three minutes the city looks strangely intact — and then the shockwave arrives, flattening buildings ring by ring outward from the impact, followed by hurricane-force winds and a rain of debris. Beyond the blast zone, broken windows injure people kilometres away, exactly as they did at Chelyabinsk. Drop the City killer preset above onto your own city to watch it happen.

The asteroids we're actually watching

There is no known asteroid on a certain collision course with Earth. But astronomers track thousands, and a handful are worth knowing by name — partly because their odds keep changing. When an object is first spotted, its orbit is uncertain, so the impact probability often rises as the error bar sweeps across Earth, then collapses to zero once more observations shrink it.

2024 YR4 — discovered in December 2024, it briefly carried the highest impact probability ever recorded for an object its size, peaking near 3% for 22 December 2032 and reaching level 3 on the Torino scale. More observations then dropped the Earth-impact odds to about 0.004% — effectively zero — though a small chance of it striking the Moon in 2032 remains. At roughly 55–65 m across it is the textbook example of how odds spike and then vanish.
Apophis — about 370 m wide. On 13 April 2029 it will pass just ~32,000 km above the surface, closer than our geostationary satellites and bright enough to see with the naked eye, yet impact has been ruled out for at least a century. Try the Apophis preset to see what that near-miss would have meant.
Bennu — sampled by NASA's OSIRIS-REx, which returned material to Earth in September 2023. It carries a roughly 1-in-2,700 chance of impact around the year 2182. Run the Bennu preset for the scale.

Figures current as of August 2026.

Could we stop one?

Yes — provided we see it coming. In 2022 NASA's DART spacecraft deliberately crashed into the moonlet Dimorphos and shortened its orbit by about 32 minutes, against a predicted minimum of 10 — the first time humanity has measurably moved a celestial body. A kinetic impactor like DART is the front-line option; slower methods such as a "gravity tractor" or an ion-beam shepherd could nudge an orbit over years, and a nuclear stand-off blast is the last resort, not the first. The single most important variable is not the size of the push but the warning time: a gentle nudge a decade out beats a violent one delivered a year before impact.

How big would an asteroid have to be to destroy a city?

These are the rough thresholds the simulator's own physics produces — smaller objects burn up harmlessly, and each step up in size is far rarer than the last.

Diameter	What happens	How often
5 m	Bright fireball; nothing reaches the ground	Yearly
20 m	Chelyabinsk — windows shatter across a city	~60 years
50 m	Tunguska — a forest or small city flattened	~500 years
140 m	A metropolitan area destroyed	~20,000 years
1 km	Continental damage and global climate effects	~500,000 years
10 km	Mass extinction	~100 million years

Asteroid impact vs a nuclear weapon

The simulator reports every impact in Hiroshima bombs for a reason: it is the only yardstick most people have. But asteroids quickly outrun our entire arsenal. The largest weapon ever detonated, the 50-megaton Tsar Bomba, is matched by a stony asteroid only around a hundred metres across — and every size above that has no human-made equivalent at all. A kilometre-wide impact releases more energy than every nuclear weapon on Earth combined, and its lasting danger is not the blast but the dust and sulphate aerosols lofted into the stratosphere — the same mechanism behind a nuclear winter or the global cooling that follows a supervolcano.

Asteroid Impact Simulator

What this tool is actually doing

The single most important question: does it reach the ground?

What each ring means

Why the casualty figure is a range, and a wide one

The things this model does not do

How likely is any of this?

Sources

What would happen if an asteroid hit Earth?

The asteroids we're actually watching

Could we stop one?

How big would an asteroid have to be to destroy a city?

Asteroid impact vs a nuclear weapon

Asteroid impact FAQ

Built from scratch by AstroAli