Drifting through the Milky Way may be billions — perhaps even trillions — of rogue planets: worlds with no sun of their own, some flung from the systems where they formed, now wandering the galaxy in darkness.

Drifting through the Milky Way may be billions, and on some estimates trillions, of rogue planets: worlds bound to no star, some flung out of the systems where they formed, others perhaps never attached to a star at all. They are also called free-floating, nomad, or starless planets. The reason the figure spans such a wide range is simple. We have confirmed very few of them, and counting things you cannot see is hard.

Microlensing searches have produced fewer than ten likely low-mass free-floating planet detections, according to a survey of the field by IEEE Spectrum. Direct-imaging surveys of young star-forming regions have turned up many more planetary-mass candidates, though their formation history is harder to classify. Almost everything beyond that handful is inference and modelling, and the gap between a small number of firm detections and a projected population in the billions is the real state of play.

How you find a planet that gives off no light

A planet with no sun emits almost nothing a telescope can catch directly. The main way these objects are found instead is gravitational microlensing, which does not see the planet at all. It sees what the planet’s gravity does to the light of a star behind it.

When a rogue planet passes almost exactly between Earth and a distant background star, its gravity bends and briefly magnifies that star’s light. The star appears to brighten, then fade, and for a planet-sized lens the whole event is short. The first credible detection of an Earth-mass candidate came this way: an event designated OGLE-2016-BLG-1928, reported in 2020 by Przemek Mróz of the University of Warsaw and colleagues, with a brightening timescale of about 41.5 minutes, the shortest then recorded and only caught because the data were taken at high cadence. The object was estimated at roughly 0.3 Earth masses. It remains a candidate rather than a confirmed rogue, because the data cannot rule out that it sits very far from a host star rather than being truly unbound.

The strength of the method is that it can find low-mass objects nothing else can reach. The weakness is built into it. Each event happens once and never repeats, which makes the mass of any single object hard to pin down, and makes the whole population something you reconstruct statistically rather than observe one by one.

Where the large numbers come from

The headline estimates rest mainly on this statistical reconstruction. A 2023 analysis of a nine-year survey by the Microlensing Observations in Astrophysics collaboration, led by Takahiro Sumi at Osaka University, concluded that free-floating planets are far more common than earlier work assumed, and supported the idea that the galaxy holds more such objects than it has stars.

It is worth being clear about what that is. It is not a tally. It is an extrapolation from a small number of brief lensing events, folded through models of how often such events should occur. One NASA-backed estimate puts the Milky Way’s rogue planets at roughly twenty per star, or trillions of worlds in total. Other modelling work lands much lower, closer to one free-floating planet per star over the mass range considered. The honest conclusion is not that anyone has counted them. It is that the population could be enormous, and the uncertainty is still enormous too.

Where they come from, which is also unsettled

There are two broad accounts of how a planet ends up with no star, and they are not exclusive.

One is ejection. A planet forms in orbit around a star, then gets thrown out, usually by a gravitational shove from a larger planet or a passing star in a crowded cluster. Simulations of dense star-forming regions, such as work modelling the Orion Trapezium cluster at Leiden, produce rogue planets this way in large numbers.

The other is that some of these objects never had a star. They may have formed directly from collapsing gas, the way stars do, but with too little mass to ignite. The International Astronomical Union has suggested the term sub-brown dwarf for objects formed like stars but below the mass needed for fusion, which blurs the line between a small failed star and a large free-floating planet.

A recent wrinkle came from the James Webb Space Telescope. Observing the Orion Nebula, Samuel Pearson and Mark McCaughrean reported around 540 planetary-mass candidates, of which about nine per cent appeared to be in wide pairs, which they nicknamed JuMBOs, for Jupiter-mass binary objects. Pairs are awkward for the ejection story, since it is hard to throw two objects out of a system together and keep them bound to each other. The result is strange enough that it remains contested, with later analyses questioning how many of the pairs hold up and whether the objects should be called planets at all.

What the next survey is built to settle

The instrument expected to move this from extrapolation toward a census is NASA’s Nancy Grace Roman Space Telescope, now targeted for launch no earlier than September 2026, with a commitment to launch no later than May 2027. Roman will run a dedicated microlensing survey from space, staring at a strip of sky toward the galactic centre for months at a time, above the atmospheric blurring that limits ground-based work.

The expectations have grown as the modelling has improved. An earlier estimate put Roman’s likely haul at around 50 Earth-mass rogue planets. A 2023 study led by Naoki Koshimoto at Osaka University raised that to roughly 400. Japan’s PRIME telescope in South Africa is intended to make simultaneous ground-based observations, which would help measure masses rather than just count events. A 2025 paper by William DeRocco and colleagues, posted to the arXiv, works through how Roman’s data could be used to reconstruct the free-floating planet mass function, the distribution of how many of these worlds exist at each mass.

There is also the prospect of pairing Roman with the European Space Agency’s Euclid, already in orbit. A joint survey, according to BBC Science Focus, could turn up more than 100 rogue planets in its first year.

What to watch

The useful thing about Roman’s survey is that it has a clear way to be wrong. The models predict hundreds of detections. If Roman finds far fewer, or none, the population estimates and the detection methods both come back under review.

For now the honest position is that rogue planets are real, that a handful have been firmly detected, and that whether they number in the billions or the trillions, and how they mostly form, are open questions a single mission is now built to narrow. The figure to watch is not the trillion. It is the first few hundred, and whether they show up.