Milky Way–Andromeda merger odds shift as new data refines collision forecast

For decades, astronomers have told a story with a date attached: the Milky Way and Andromeda will collide. In 2012, NASA announced that Hubble observations pointed to a nearly head-on encounter roughly four billion years in the future, with the two galaxies taking another two billion years after that to fully merge into a single system. The narrative was clean, the timeline was concrete, and it entered the public imagination as settled fact.

Then the measurements changed. In 2019, the European Space Agency released new data from the Gaia satellite, which had tracked the movements of thousands of stars across both galaxies with unprecedented precision. That work, led by Roeland van der Marel, suggested a less direct collision and pushed the first close approach back to about 4.5 billion years. The forecast had shifted, but it still carried the weight of inevitability. Andromeda was still coming.

In June 2025, Till Sawala and his team published a paper in Nature Astronomy that upended the conversation entirely. Using both Hubble and Gaia observations, they modeled not just the Milky Way and Andromeda, but also two smaller companion galaxies: Triangulum and the Large Magellanic Cloud. When they varied their starting measurements within the range of uncertainty—rather than following a single preferred trajectory—they found something unexpected. The odds of a merger within the next ten billion years came out to roughly fifty-fifty. The collision was no longer assured.

The shift reveals something fundamental about how we predict the future of the cosmos: the forecast is only as solid as the measurements that feed it. Andromeda lies 2.5 million light-years away. To the naked eye, it appears as a faint patch. Measuring its motion across the sky requires identifying individual stars within it and tracking their tiny displacements—a task of extraordinary precision. The relative speed at which the two galaxies approach each other, roughly 400,000 kilometers per hour, is straightforward enough to calculate. But the sideways motion—how far to one side Andromeda will pass—is far harder to pin down. That uncertainty cascades through the entire calculation.

The smaller galaxies matter more than their size might suggest. Triangulum, in Sawala's models, actually increased the likelihood of a merger. The Large Magellanic Cloud pushed in the opposite direction, altering the Milky Way's motion and introducing movement outside the original orbital plane. These are not merely extra mass to be added to the equation; their position and trajectory reshape the gravitational dance itself. Treating them as passive participants misses the point entirely.

The 2025 finding also clarified a distinction that had been blurred in earlier headlines. A first close approach and a completed merger are different events separated by billions of years. In some of Sawala's simulations, the galaxies passed within roughly half a million light-years of each other, moved apart, and then returned for a later merger. The extended dark matter halos surrounding each galaxy interacted through a process called dynamical friction, transferring energy out of the orbit and allowing the separation to shrink over successive encounters. Other passages were too wide for that process to work, leaving no merger within the ten-billion-year window being examined. The 2025 calculation also found only about a two percent chance of the earlier, nearly head-on collision occurring in four to five billion years.

But the story did not end there. In April 2026, Hao Wu and colleagues revisited the problem using the same broad modeling framework but with newer Gaia-based measurements of Andromeda's and Triangulum's motions, corrected for systematic offsets. Their preferred model raised the merger probability to about ninety percent, with a median merger time of roughly 6.5 billion years. Across the range of measurement uncertainty they examined, the probability ranged from about 64.7 percent to 100 percent. Yet even Wu's team described the measurements as inconclusive. They called for proper-motion uncertainties of two microarcseconds per year or smaller to substantially narrow the probability range. That level of precision remains the next frontier.

What these successive studies demonstrate is not that astronomers are confused, but that a probability forecast must always travel with its assumptions. Revised inputs produce different results without resolving every uncertainty. The Milky Way and Andromeda may merge, or they may not. The odds have shifted from near-certain to fifty-fifty to ninety percent in the span of a few years, not because the galaxies changed their course, but because our ability to measure that course remains imperfect. The forecast will continue to evolve as the measurements improve.