Red Light Works, But How Well Depends on Two Numbers

Red light therapy has randomized controlled trials behind it, and they found real effects on skin. Measurably firmer skin. Shallower wrinkles. Higher collagen density. That puts it ahead of nearly everything else sold for your face. Here’s the catch: the effect only shows up at particular doses of light, and whether the device sitting in front of you delivers one comes down to two numbers most manufacturers never bother printing on the box. Which means this decision doesn’t get made by the marketing copy. It gets made on the spec sheet.

The formal name is photobiomodulation: light that changes how cells behave without heating the tissue. It is not a laser. It is not a heat lamp. And it is absolutely not a tanning bed. Tanning beds emit ultraviolet light, which damages DNA in your skin. Red light devices sit at the opposite end of the visible spectrum, plus a sliver of the invisible band just past it, and carry zero ultraviolet. None. That’s the whole point.

OneThe Two Wavelengths That Matter

Two bands do all the work here, and the difference between them comes down to one thing: depth.

Red light, roughly 630 to 660 nanometres. Visible. It’s the colour you see glowing. It reaches the epidermis and the upper dermis, which is exactly where most of what you see in the mirror gets decided.

Near-infrared, roughly 830 to 850 nanometres. Just past what your eye can register, so a panel running it at full output can look like it’s barely doing anything. It travels deeper into you, which is exactly why devices built for joints and muscle lean on it instead.

Trials that combine both wavelengths tend to beat either one alone on skin outcomes. So a device offering both is doing something the research backs up, not just padding a spec list.

TwoWhat’s Happening in the Cell

The leading explanation involves an enzyme called cytochrome c oxidase, which sits inside the mitochondria, the compartments inside your cells that turn food and oxygen into usable energy. That enzyme absorbs red and near-infrared light. Here’s the proposed chain: absorbing that light nudges the enzyme to work more efficiently. The cell makes more ATP, the molecule cells spend when they need energy. And a cell running on more energy does more of its job. In skin, the cells whose job matters most for how you look are fibroblasts, the ones that manufacture collagen and elastin.

That chain is plausible, and it has laboratory support at each individual link. It is not settled science, not yet. The effect shows up in trials whether or not anyone can fully explain why, and the chain above is simply the best explanation anybody’s got right now.

That’s a completely normal situation in medicine, more normal than people assume. Aspirin worked for seventy years before anyone figured out what it was doing to prostaglandins.

ThreeThe Numbers From the Trials

Skin is where this evidence is at its strongest. Here are the numbers.

A randomized controlled trial measuring wrinkle volume around the eyes found roughly a 30 percent reduction after a full course of treatment. Thirty percent, measured, not eyeballed. A split-face trial in 137 women aged 40 to 65 put red light on one side of the face and amber light on the other, ten sessions across four weeks. That design makes every participant her own control, so the comparison never gets muddied by one person’s skin just being different from another’s. Trials measuring collagen density in the dermis with ultrasound found real increases against control too.

Firmer skin. Smoother texture. Better tone. All of it over one to three months of consistent use, measured with instruments against a control, not judged by somebody squinting in bad bathroom light. For twelve minutes a day with zero downtime and zero irritation, that’s a genuinely good return. Red light produces a smaller change than a retinoid does. It also isn’t a procedure. The fair comparison isn’t a retinoid. It’s doing nothing at all. Against nothing, red light wins clean.

The change is real. It’s measurable. It takes one to three months, not three days. It behaves like a routine, not a treatment, which is exactly why it’s easy to keep up with.

FourThe Two Numbers That Decide Everything

Two numbers decide whether a device does anything at all for you. Irradiance is how much light power lands on a given area, written in milliwatts per square centimetre. Fluence is the total dose delivered, irradiance multiplied by time, written in joules per square centimetre. The trials that found real effects generally used irradiance somewhere between 20 and 60 mW/cm² and delivered 4 to 18 J/cm² per session.

Irradiance falls off hard with distance. A panel measured at six inches is not delivering the same dose at eighteen inches, not even close, and the drop happens a lot faster than intuition tells you it should. Which is exactly why “hold it near your face for ten minutes” is not a protocol. It’s a shrug. Without an irradiance figure tied to a stated distance, you cannot calculate what you’re getting. Neither can the person who sold it to you, and that should tell you something.

A device that will not publish its irradiance at a specified distance has not been shown to sit anywhere near the range where an effect was ever demonstrated. It might sit right in that range. It might not. Without the figure, neither you nor the seller can tell, and one of you is the one who paid for it.

FiveMore Light Is Not Automatically Better

Photobiomodulation has what researchers call a biphasic dose response, which is a fancy way of saying more is not automatically better. Going up from nothing, the effect climbs. Past a certain point, it drops back down again. Push past that, and it can do harm. Too much light appears to generate enough reactive molecules to overwhelm your cell’s own defences, which is the exact opposite of what you were going for.

That’s unusual, and worth sitting with for a second. Most things you buy for your skin get slowly better with more use, or they plateau and just sit there. This one’s different: doubling your session length can shove you straight past the peak and back down the other side of it. Too much light leaves you worse off than the correct dose ever would have.

So here’s the instruction that matters most, and it’s also the least satisfying one. Follow the protocol. Do not add sessions because you’re impatient for faster results. Treat the manufacturer’s stated time as a hard ceiling, never a suggestion.

SixWhat to Check Before You Buy Anything

A 2025 review of this entire product category went and counted how many devices had been studied. Of 148 products listed in the relevant FDA database, two, just two, had an associated clinical trial. Two out of 148 had been studied. At all.

So the questions you need to ask are narrow, and they’re specific.

Does it state its wavelengths? You want real numbers, 660 nm, 850 nm, not vague marketing phrases like “red light technology” or “clinical spectrum”.

Does it state irradiance at a stated distance? Something like 45 mW/cm² at six inches. One number without the other one is useless to you. This single question alone eliminates most of the market, immediately.

Is it FDA-cleared, and can you verify that? Clearance is a lower bar than approval, and it mostly concerns safety rather than whether the thing works. But it’s checkable in a public database. A clearance claim that doesn’t show up there tells you something, and it’s not a good something.

Does the manufacturer link to published research on its own device, or just to research on the general idea of red light? Most link to the second one. The 148-to-2 number from a minute ago tells you exactly why.

One note on the safety side, and it’s a real one. A well-known FDA-cleared LED face mask got pulled from the market over a theoretical risk of eye injury, and that happened despite the thing having clinical support behind it. Clearance is not a permanent verdict. It can change.

SevenProtect Your Eyes

Near-infrared is invisible, and your blink reflex only responds to light you can see. That’s the specific hazard here: at full output, a near-infrared emitter can look like it’s barely even on, and nothing in your body is going to tell you to look away.

Wear the goggles that came with the device. Or keep your eyes closed for facial treatments, either one works. Do not look directly into a panel to check whether it’s working. Just don’t. Handled sensibly, this is a small risk. It’s also the one part of this whole practice where the downside is bigger than twelve wasted minutes.

Where it fitsSunscreen and Retinoids Still Come First

If you’re building a routine, the order matters more than the device.

Daily broad-spectrum sunscreen prevents the exact damage everything else on this list is trying to repair, and it has more evidence behind it than every other step here combined. A retinoid at night is the single best-evidenced ingredient for aging skin, full stop. Both of those beat red light on evidence, on effect size, and on cost, by a wide margin, no contest.

Red light earns its place once those two are locked in and consistent, not before. It works alongside them. Not instead of them. Our skincare library sets it next to every other step with the exact same grading applied, so you can see precisely where it sits in the lineup.

It does something real, on a slow timeline, provided the device in front of you produces the dose the studies used. Most of this decision comes down to checking whether it can. And a device that publishes its numbers is, almost always, a device that has them worth publishing.

Yojo Dojo may earn from some links on this site. We grade every technique on the same evidence scale whether or not we recommend a product for it. Educational content only, not medical advice. Speak with a dermatologist about your own skin, particularly if you have a photosensitive condition or take a medication that increases light sensitivity.