Time from today 0.00 Gyr
MW–M31 separation 766 kpc
Relative speed 110 km/s
Sun–MW centre 8.2 kpc
M31 size in our sky 3.0°
Closest approach –
Approach
Outside view
Earth night sky
Fit view
Moon 0.5°
Press Play (or drag the timeline) to run the next 10 billion years. Drag to rotate, wheel or pinch to zoom, tap a star to follow it.
Computing…
⏮
▶ Play
0.00 Gyr
0.02 Gyr/s 0.05 Gyr/s 0.12 Gyr/s 0.3 Gyr/s 0.7 Gyr/s
Andromeda–Milky Way Collision
Milkomeda: a restricted N-body (Toomre-type) model. Educational model, not a forecast.
Scenarios
Standard (HST 2012)near head-on, merges ~7 Gyr
Gaia glancingwide pass, no merger in 10 Gyr
Head-on, coplanardiscs face-on, dramatic tails
With M33Triangulum joins the dance
M31 onlyonly Andromeda's stars; MW is the perturber
Parameters
M31 sideways speed
M31 / Milky Way mass ratio
Milky Way disc tilt
M31 disc tilt
M31 disc spin angle
Quality Low (6,000) High (16,000) Focus System Milky Way Andromeda Tracked star
Earth-sky zoom
Graphs
Where does the Sun end up?
Re-roll the Sun's position Back to the Sun
160 Sun-like tracers start on a ring at 8.2 kpc around the Milky Way centre (different azimuths). The fractions are one run of a simplified model, not a probability estimate. Cox & Loeb (2008) found roughly even odds that the Sun ends up farther from the merged centre than it is today.
Do stars collide?
Stellar density (stars per pc³)
Stars in the colliding region
The science
Setup
Today (t = 0) M31 is about 766 kpc (2.5 million light-years) away and approaching at about 110 km/s (radial velocity from spectra; Hubble proper motion by van der Marel et al. 2012). The sideways speed vt is the uncertain part: van der Marel et al. (2012) found a nearly radial orbit (vt of order 20 km/s, within errors up to ~57 km/s), giving a first close pass in roughly 4 to 5 billion years and a final merger around 6 to 10 billion years from now. Gaia-based revisions (for example Salomon et al. 2021) find more sideways motion, and Sawala et al. (2025) put the chance of a merger within 10 Gyr near 50%. The "Gaia glancing" scenario shows the wide-pass outcome. All numbers here are approximate.
Restricted three-body (Toomre) approach
The two galaxy centres are integrated as extended point masses. The stars are massless test particles that feel the gravity of both centres but do not pull on anything (Toomre & Toomre 1972). Each galaxy is a halo + bulge + disc:
Hernquist halo/bulge: M(r) = M r² / (r + a)²
a(r) = - G M(r) r̂ / r² (force on a star)
disc (sphericalised Plummer): M(r) = M r³ / (r² + b²)^(3/2)
circular speed: v_c = sqrt( G M(<r) / r )
with G = 4.30091×10-6 kpc (km/s)2 / M☉ . Disc stars start on near-circular orbits with an exponential surface density (scale length 3.5 kpc for the Milky Way, 5.3 kpc for M31); bulge stars are a roughly isotropic sphere. Orbits use a drift-kick-drift leapfrog with steps of at most 2 Myr.
Gravity and friction between the centres
F = G M₁(<s) M₂(<s) / s² (s = softened separation)
adf = - 4π G² M ρ lnΛ [erf(X) - (2X/√π) e-X² ] v̂ / v², X = v / (√2 σ)
The second line is Chandrasekhar's (1943) dynamical friction: each halo drags on the other, which is what turns a flyby into a merger. The friction strength was tuned so the standard scenario gives a first pass near 4 Gyr and a merger near 7 Gyr, in line with published full simulations. When the cores come within 4 kpc and are slow they are treated as merged and settle together.
Stars almost never collide
P = n σ L, σ = π (2 R☉)², R☉ = 2.25×10-8 pc
n is the stellar number density and L the path length through the other disc. The mean spacing between stars (about 2 pc near the Sun) is tens of millions of stellar diameters, so a whole merger yields at most a handful of physical collisions. The calculator above uses your numbers.
What happens to the Sun and Earth
Cox & Loeb (2008) estimate the Sun's orbit is changed but the Solar System is not disrupted: about even odds that it ends farther from the merged centre than today, and a small chance of being thrown into a tail. Earth is not directly threatened. The Sun's slow brightening, which makes Earth too hot for liquid oceans in roughly a billion years, matters much sooner than the merger.
What the "view from Earth" shows
The camera sits on the tracked star and looks at M31. M31's disc radius is taken as 20 kpc, so the angular diameter is θ = 2 atan(20 kpc / d). Today d = 766 kpc gives about 3 degrees, six Moon diameters. Brightness is exaggerated: to the naked eye only the core is visible. M31's real disc is inclined about 77 degrees to our line of sight, which the default tilt and spin reproduce.
Uncertain or simplified
Static spherical halos (no live halo response); massless stars (no bars, spiral arms or self-gravity); no gas, star formation or central black holes; dark-energy expansion ignored; simplified friction; M33 placed illustratively; the Milky Way and M31 masses (about 1.0×1012 and 1.4×1012 M☉ ) are uncertain by tens of percent; the orientation of the discs relative to the orbit is poorly known; and the tidal tails and final shape are one realisation of an approximation. Treat timings as order-of-magnitude.