Milky Way and Andromeda Collision Simulation

Andromeda, the nearest large galaxy to ours, is about 2.5 million light-years away and closing in at roughly 110 km/s, close to 400,000 km/h. For years the standard story was simple: in about 4 billion years the two galaxies collide, and a few billion years later they merge into one. Newer measurements say the future is far less certain. This simulation lets you watch three possible outcomes, from today to 10 billion years from now.

Drag to rotate the view, use the buttons to zoom, and switch between scenarios at any time; the timeline keeps its place, so you can compare the same moment in each future.

Will the Milky Way and Andromeda collide?

Maybe. Andromeda is definitely heading towards us; its light is shifted towards blue by its approach. What decides the outcome is how fast it is also drifting sideways, and that motion is very hard to measure from 2.5 million light-years away.

• 2012: Hubble measurements suggested almost no sideways motion, so the galaxies would meet nearly head-on in about 4 billion years and merge about 2 billion years after that.
• 2019: Gaia data pointed to more sideways motion, which pushes the first encounter later and makes it less direct.
• 2025: A study combining Hubble and Gaia data, led by Till Sawala and published in Nature Astronomy, ran thousands of simulations across the measurement uncertainties. It found roughly even odds of a merger within the next 10 billion years. Part of the change comes from including the pull of the Large Magellanic Cloud, a satellite galaxy of the Milky Way.

The three scenarios in the simulation are illustrative paths across that range of possibilities, not predictions:

• Direct hit: a nearly head-on encounter, as the 2012 measurements suggested. A first pass in a bit over 4 billion years, then a merger about a billion years later.
• Close pass, later merger: a wider first pass that throws out long tidal tails, a long swing back out, and a merger nearer 8 billion years.
• Near miss: a faster sideways drift. The galaxies pass hundreds of thousands of light-years apart and don’t merge within 10 billion years.

When will Andromeda collide with the Milky Way?

If it happens, the first close encounter is most likely 4 to 5 billion years from now. That part of the timeline is fairly robust, because it depends mostly on Andromeda’s well-measured approach speed. Whether that first encounter is a direct hit, a glancing pass or a miss, and whether a merger follows, is what remains uncertain.

If the galaxies do merge, the result will be a single large, roughly elliptical galaxy, sometimes nicknamed Milkomeda. In the direct-hit and close-pass scenarios you can watch the two spiral disks being torn into tails and arcs, then settling into one round, featureless swarm of stars.

What happens to the Sun and Earth?

Almost certainly nothing violent. Galaxies are mostly empty space: the nearest star to the Sun is more than 4 light-years away, while the Sun is less than 5 light-seconds across. When two galaxies collide, their stars pass between each other, and direct star-to-star collisions are vanishingly rare. What changes is the Sun’s orbit. In many simulations, including the ones here, the Sun is flung to a different distance from the new galactic centre, often farther out. The panel under the simulation tracks the Sun’s distance from the centre of mass as time runs.

The timing has a twist, though. In about 5 billion years the Sun itself will swell into a red giant, and Earth will become uninhabitable long before that as the Sun slowly brightens. So no one on Earth is likely to watch the collision.

What will the night sky look like?

Switch to Night sky from the Sun to see the view from the star standing in for the Sun, looking towards Andromeda. Today Andromeda spans about 3°, roughly six full Moons, although only its bright core is easy to see without a telescope. As the galaxies approach it grows across the sky, and around the first encounter it fills a large part of it, crossing the band of the Milky Way. After a merger, the band of our own disk is replaced by the diffuse glow of the merged galaxy.

How this simulation works

A full simulation of colliding galaxies, where every star pulls on every other star, takes a supercomputer. This one uses a much cheaper method, the one Alar and Juri Toomre used in 1972 to show how tidal tails form:

• Only the two galaxy cores have gravity. Each galaxy is modelled as a massive halo of dark matter (a Hernquist profile), which holds most of a galaxy’s mass. The Milky Way is set to 1.0 trillion and Andromeda to 1.3 trillion solar masses, within the published range. The two halos pull on each other, and dynamical friction, the drag a moving mass feels as it plows through the other halo, lets them lose energy and merge.
• The stars are test particles. Each one orbits in the combined pull of both halos but does not pull on the other stars. That lets about 100,000 of them (about 33,000 on phones) be moved in parallel on your graphics card, at full frame rate.
• The starting conditions are today’s real values: Andromeda about 785 kiloparsecs away and approaching at 110 km/s. Only the uncertain sideways speed differs between the scenarios.

It is a model, not a research simulation. It leaves out gas and new star formation, the two galaxies’ disk orientations are illustrative, and the merged galaxy’s final shape is approximate. What it does capture is the part that matters here: how the galaxies’ orbits and tidal forces shape the encounter, and how much the outcome depends on a sideways speed of a few dozen kilometres per second.

For a sense of scale on a different extreme, see the TON 618 size comparison, which puts the largest known black holes next to the Solar System.

Sources

• Andromeda’s distance and approach: van der Marel et al. (2012, 2019), from Hubble and Gaia proper motions.
• 2025 merger odds: Sawala et al., Nature Astronomy (2025); see the ESA and ESA/Hubble releases.
• Method: Toomre & Toomre (1972), restricted three-body simulations of tidal tails; Chandrasekhar (1943), dynamical friction.