How We Find Exoplanets — and Why We've Barely Seen Any

As of August 2026, astronomers have confirmed more than 6,000 planets orbiting other stars. We have taken a direct picture of about 98 of them.

That gap is the most important thing to understand about exoplanets. Almost everything we know about worlds beyond our solar system comes not from looking at the planets, but from watching their stars very carefully and inferring that something must be tugging at them or crossing in front of them. We are reading shadows and wobbles, not photographs.

This has a consequence that rarely makes the headlines. Each detection method is good at finding one kind of planet and nearly blind to others — so the catalogue of 6,000 worlds is not a fair sample of what is out there. It is a portrait of what our particular techniques happen to notice.

Here is how the methods actually work, and how to read the resulting map of the galaxy without being misled by it.

Six Thousand Worlds, Almost None of Them Seen

The NASA Exoplanet Archive tabulates every confirmed exoplanet by the method that first found it. The distribution is extraordinarily lopsided:

Two methods — transit and radial velocity — account for 92.7% of every planet we know about. Both are indirect. Neither sees the planet at all.

Direct imaging, the only technique that actually collects light from the planet itself, has produced 1.5% of the total. When you see a vivid picture of an alien world, it is almost always an artist's impression, including the one at the top of this article.

The Transit Method: Watching for a Shadow

The workhorse is beautifully simple. If a planet's orbit happens to be edge-on from our point of view, the planet passes in front of its star once per orbit and blocks a sliver of light. Measure the star's brightness precisely enough and you see a tiny, repeating dip.

The dip is small. An Earth-sized planet crossing a Sun-sized star blocks roughly 0.008% of its light — the equivalent of noticing one dimmed bulb in a stadium. Doing this reliably is why space telescopes matter: Earth's atmosphere flickers far more than the signal does.

What a transit gives you is unusually rich for an indirect measurement:

• Orbital period, from the gap between dips
• Planet size, from the depth of the dip — a bigger planet blocks more light
• Orbital distance, derived from the period and the star's mass
• Atmospheric composition, if you can catch starlight filtering through the planet's atmosphere on the way past

That last point is why NASA describes transit spectroscopy as reading a barcode: molecules absorb specific wavelengths, so starlight that has passed through an atmosphere carries a chemical fingerprint.

Radial Velocity: Listening for the Wobble

The second method is older and cleverer. A planet does not simply orbit its star — both orbit a shared centre of mass. So the star traces a small circle too, moving slightly toward us and then slightly away.

That motion shifts the star's spectrum: its light is squeezed towards the blue as it approaches and stretched towards the red as it recedes, exactly as starlight's wave nature predicts. Measure that shift over time and you can infer an unseen companion.

This is how the field began in earnest. In October 1995, Michel Mayor and Didier Queloz announced 51 Pegasi b — the first planet found around a Sun-like star, at least half Jupiter's mass, whipping around its star every 4.2 days. Nobody had expected a giant planet that close in, and the discovery won the 2019 Nobel Prize in Physics.

Radial velocity gives you the orbital period and a minimum mass. The minimum matters: unless you know the orbit's tilt, you cannot tell a modest planet seen edge-on from a heavier one seen at an angle. Transit and radial velocity are therefore complementary — one gives size, the other gives mass, and together they give density, which is how you distinguish a rocky world from a puffball of gas.

The Geometry Problem Nobody Mentions

Here is the constraint that quietly shapes the entire catalogue.

The transit method only works if the orbit is lined up edge-on from our vantage point. Most are not. The geometric chance that a randomly oriented orbit produces a transit is roughly the star's radius divided by the orbital distance — and that number is brutal for anything far from its star.

Read the middle row again. If there were an exact copy of Earth orbiting an exact copy of the Sun, there is a 99.5% chance we could never detect it by transit, no matter how good our telescopes became. The alignment simply is not there.

This is not a limitation we can engineer away. It is a fact about viewing angles. Every transit survey is therefore sampling a thin, biased slice of the planets that exist.

Microlensing and Direct Imaging: The Other 6%

Two further methods fill in gaps the big two cannot reach.

Gravitational microlensing (282 planets) uses relativity. When one star passes precisely in front of a more distant one, its gravity bends and magnifies the background star's light. If the foreground star has a planet, the planet adds a brief extra spike to that brightening. Microlensing is uniquely good at finding planets far from their stars — and even planets bound to no star at all — but it has a hard drawback: the alignment never repeats. You get one look, and the system is gone.

Direct imaging (98 planets) does what everyone imagines astronomy to be: blocking the star's overwhelming glare with a coronagraph and collecting the planet's own photons. It is spectacularly difficult, because a star can be a billion times brighter than the planet beside it. NASA notes that such images have been "mainly confined to giant planets still so hot from their fresh creation that they remain self-luminous" — in other words, young, enormous, and far from their star.

So the one method that truly sees planets is restricted to the least Earth-like planets imaginable.

Why Our Catalogue Is a Distorted Sample

Put the biases side by side and the shape of the problem becomes clear. Each method has a sweet spot, and the sweet spots barely overlap where we most want to look.

The practical implications are worth stating plainly:

• Early results were dominated by hot Jupiters — massive planets on tight orbits — because those are the easiest thing to find by both dominant methods. NASA notes this bias explicitly in its own account of early radial-velocity detections. That was never evidence that hot Jupiters are typical.
• Small planets on wide orbits are the hardest case for every method we have. That includes Earth.
• When a study says a planet type is "common," check the method. Frequency estimates have to be corrected for detection efficiency, and those corrections are large, model-dependent, and where much of the real scientific argument happens.

None of this makes the catalogue untrustworthy. It makes it a measurement with a known instrument response — which is a normal state of affairs in observational science, and the same care needed when reading claims about the universe's large-scale behaviour.

What JWST Changed: From Counting to Characterising

For twenty years the central question was how many. Kepler answered it emphatically: planets are ordinary, and most stars have them. The question has now shifted to what are they like, and that requires reading atmospheres rather than counting dips.

This is where the James Webb Space Telescope reset the field. Its infrared sensitivity lets astronomers take the faint transit-spectroscopy signal — starlight filtered through a planet's atmosphere for a few hours — and pull real molecular detections out of it. The work is painstaking and the debates are genuine, because the signals sit close to the noise and stellar activity can mimic a planetary feature.

The honest position on biosignatures is worth stating clearly: detecting a molecule is not detecting life. A gas that living things produce on Earth can often be produced by geology or photochemistry elsewhere, so any claim of life needs the alternatives ruled out — and that is a much higher bar than a detection.

How We Got Here

The pulsar planets at the top of that list are the ones illustrated in this article's hero image — three worlds orbiting the collapsed core of an exploded star, bathed in radiation. They were found by timing the pulsar's beat and noticing it arrived slightly early and slightly late. Even the very first exoplanets were detected by watching something else.

Common Misconceptions

"We have photographs of exoplanets." For about 98 of them, in the sense of a few pixels of genuine planetary light. For the other 98%, no. Colourful planetary landscapes are artists' impressions, and reputable outlets label them as such.

"Better telescopes will let us see them all." Bigger telescopes help enormously with imaging and spectroscopy, but they cannot fix the transit alignment problem. If an orbit is not edge-on to us, no instrument makes it transit.

"Most planets are hot Jupiters." Most early discoveries were. That was a detection artefact, and the correction is one of the field's clearest lessons in reading its own instruments.

"Finding a molecule means finding life." No. Non-biological chemistry can produce many of the same gases, so a biosignature claim requires eliminating those routes — which is far harder than the detection itself.

The Bottom Line

We have found more than 6,000 planets around other stars, and we did it almost entirely by watching stars rather than planets. Transit and radial velocity between them account for 93% of the catalogue; direct imaging, the only method that sees a planet's own light, accounts for 1.5%.

The result is a rich map with a known distortion. Our methods are best at large planets close to their stars and worst at small planets in temperate orbits — which is to say, worst at planets like ours. Keeping that in mind is the difference between reading the exoplanet catalogue as a census of the galaxy and reading it, correctly, as a record of what our instruments can currently notice.

Explore more in our space and astronomy hub, or start with the physics of the starlight all of this depends on in what is a photon.

Frequently Asked Questions

How many exoplanets have been discovered?

More than 6,000 are confirmed as of August 2026, with thousands of additional candidates awaiting confirmation. The NASA Exoplanet Archive's detailed breakdown lists 6,336 confirmed planets sorted by the method that found them. The number climbs steadily, so any figure you read is a snapshot — the rate of discovery has accelerated sharply since the first detection just over three decades ago.

What is the most successful way to find exoplanets?

The transit method, by a wide margin: 4,676 planets, or about 74% of the total. It works by measuring the tiny dip in a star's brightness when a planet crosses in front of it. Radial velocity is second with 1,197 planets, detecting the star's slight motion toward and away from us. Together these two indirect methods account for roughly 93% of all confirmed exoplanets.

Have we ever actually photographed an exoplanet?

Yes, but rarely — about 98 planets, roughly 1.5% of the catalogue, have been directly imaged. It requires blocking the star's light with a coronagraph, since a star can outshine its planet by a factor of a billion. The successes are heavily skewed toward young, very large planets on wide orbits that are still hot enough to glow on their own. Almost every colourful exoplanet picture you have seen is an artist's impression.

Could we detect another Earth?

It is extremely difficult, for a reason that has nothing to do with telescope quality. Orbital geometry gives an Earth analogue only about a 0.47% chance of transiting from our viewing angle, compared with roughly 9.3% for a planet on a very close orbit. An Earth-mass planet also exerts a far weaker gravitational tug than a gas giant. Small planets in temperate orbits sit in the blind spot of every method we currently have.

What was the first exoplanet ever found?

Two answers, both correct. The first confirmed exoplanets were announced in 1992 by Aleksander Wolszczan and Dale Frail — two super-Earth-mass planets orbiting the pulsar PSR B1257+12, found by timing irregularities in the pulsar's beat. The first planet around a Sun-like star came in October 1995: 51 Pegasi b, discovered by Michel Mayor and Didier Queloz, who received the 2019 Nobel Prize in Physics for it.

Can we tell whether an exoplanet has life?

Not yet. JWST can now detect molecules in some exoplanet atmospheres, which is a genuine advance, but detecting a molecule is not detecting life. Many gases that living organisms produce on Earth can also be generated by geology or photochemistry, so a credible biosignature claim requires ruling out every non-biological explanation. That is a substantially harder task than the detection, and no such claim has been established.

Sources

• Exoplanet Archive: confirmed planet counts by discovery method — NASA Exoplanet Science Institute / Caltech
• Exoplanets overview — NASA Science
• How We Find and Characterize Exoplanets — NASA Science
• The Nobel Prize in Physics 2019: press release — The Royal Swedish Academy of Sciences
• Nobel Winners Changed Our Understanding with Exoplanet Discovery — NASA Science
• From one exoplanet to six thousand — Nature Astronomy (2025)