If you’re comparing devices, the red light therapy wavelength for hair growth is the first spec worth checking and the one most product pages bury under a bigger, dumber number. Diode count is the headline. Wavelength is the thing that determines whether any of those diodes are doing something. A 300-diode device emitting light your follicles can’t use is a very well-engineered lamp. Here’s what the numbers mean, which ones are supported, and where the spec sheet starts lying to you.
Two Windows, and Everything Else Is Decoration
Essentially all of the credible hair research sits in two bands:
- Red, roughly 630–680nm — usually implemented at 650, 655, 660, or 680nm. This is the visible red glow, and it’s what most FDA-cleared hair devices emit.
- Near-infrared, roughly 800–850nm — usually 808 or 810nm. Invisible. Penetrates deeper.
That’s it. If a device’s marketing leads with a wavelength outside those windows for hair specifically, you’re being sold something that was designed for a different job.
Why Those Numbers and Not Others
The mechanism has a named target, which is unusual in this industry and worth appreciating.
Inside your mitochondria sits cytochrome c oxidase, the last enzyme in the electron transport chain. It has absorption peaks that land squarely in the 630–680nm range — meaning it is unusually good at catching photons at exactly those wavelengths. When it does, it appears to release bound nitric oxide, which had been sitting there acting as a brake on cellular respiration. Take the brake off and electron transport picks up, ATP production rises, and the freed nitric oxide improves local blood flow to the follicle. Downstream, this seems to activate Wnt signalling — WNT3 and WNT10B specifically — which is one of the pathways that governs whether a follicle enters its growth phase.
That’s the theory, and it’s a decent one. It also explains the wavelength choice rather than just asserting it.
Now for why the rest of the spectrum is out. Blue and green light gets absorbed almost immediately by haemoglobin and melanin — it never reaches the follicle. Far infrared is absorbed by water, which means it converts to heat rather than signalling; that’s a warm sensation, not photobiomodulation. Between roughly 650 and 850nm there’s a gap where tissue is relatively transparent — the so-called optical window. The useful wavelengths aren’t chosen because they’re red. They’re chosen because they’re the ones that get in.
650 vs 660 vs 680: Stop Worrying About It
Manufacturers will tell you their 660nm is meaningfully superior to a competitor’s 650nm. It isn’t. Cytochrome c oxidase doesn’t have a knife-edge absorption peak at a single nanometre; it has a broad band. Ten nanometres of difference inside that band is noise. Theradome runs 680nm, Capillus runs 650nm, plenty of others sit at 655 or 660, and all of them are in the same neighbourhood doing the same thing.
If a brand’s core differentiator is five nanometres, that’s a brand without a differentiator.
Near-Infrared: The Genuinely Interesting Argument
The dermal papilla — the structure at the base of the follicle that actually controls growth — sits a few millimetres down. Red light at 660nm gets there, but attenuated. Near-infrared at 808nm penetrates deeper before scattering out, which is the reasonable case for including it.
Some devices now run both bands, and you’ll see figures quoted for dual-wavelength superiority — a commonly repeated claim is 51% greater improvement in thickness versus single-wavelength. Treat that number carefully. It circulates mostly through manufacturer and retailer content rather than independent replication, and the underlying comparisons tend to be small. The 2025 systematic review covering 38 studies and roughly 3,000 patients found LLLT beat sham across the board — including across both comb- and helmet-type devices — without establishing that dual-wavelength is a category above single.
The honest read: near-infrared is mechanistically sensible and probably a mild plus. It is not the difference between working and not working, and it does not justify a thousand-dollar premium on its own.
Laser vs LED: A Smaller Deal Than the Price Difference Suggests
Laser diodes produce coherent, tightly collimated light. LEDs produce the same wavelengths incoherently across a wider spread. The industry has spent years insisting coherence is essential, largely because laser diodes cost more.
The evidence doesn’t strongly support the premium. What matters far more is whether the right wavelength arrives at the follicle at the right intensity for long enough. That said, lasers do concentrate output, so a laser device often delivers a usable dose through hair more reliably than a diffuse LED array — the advantage is real but it’s about delivered dose, not about coherence being magic. Several FDA-cleared devices use a mix of both, which tells you the regulatory bar doesn’t consider it decisive either.
The Spec That Actually Decides Outcomes
Wavelength gets you in the door. Dose determines what happens next, and almost nobody advertises it clearly.
Two numbers matter: irradiance (how much power lands per unit area, in mW/cm²) and fluence (irradiance multiplied by time, in J/cm² — the total energy delivered). Photobiomodulation follows a biphasic dose response, which is a technical way of saying the curve is an inverted U. Too little does nothing. The right amount stimulates. Too much stimulates less than the right amount.
This has two practical consequences. First, doubling your session time to get faster results is actively counterproductive. Second, a device that claims meaningful results from a 90-second session is almost certainly underpowered, over-powered, or simply making it up — the clinically studied protocols generally run 20 to 30 minutes, several times a week.
How to Read a Spec Sheet Without Being Had
Look for:
- A stated wavelength in nanometres, in the 630–680 and/or 800–850 bands.
- A stated output — total mW, and ideally irradiance at the scalp.
- A defined protocol — session length and weekly frequency that resembles the trial protocols.
- An FDA 510(k) clearance for treating pattern hair loss specifically.
Walk away from:
- No nanometre figure anywhere. If they won’t say, there’s a reason.
- Absurdly wide ranges like “600–1000nm,” which usually means a grab-bag of cheap diodes and no design intent.
- Generic body-recovery panels repurposed for hair. They’re built to flood a torso from a distance, not to deliver a controlled dose to a scalp through a head of hair. The geometry is wrong even when the wavelength is right.
- “Infrared” with no further detail. Near-infrared at 810nm and the far-infrared in a sauna are not the same technology and don’t do the same thing. The word alone tells you nothing.
The Bottom Line
For hair, the supported red light therapy wavelengths are roughly 630–680nm and 800–850nm, and the reason is specific: cytochrome c oxidase absorbs strongly there, and tissue is relatively transparent there. Inside those bands, small differences don’t matter — 650 versus 660 is marketing. Near-infrared is a sensible addition, not a revolution. Laser versus LED matters less than the marketing implies. And the variable that actually decides your result — delivered dose over an adequate session, repeated for months — is the one that almost never makes it onto the box.
Informational only, not medical advice. Use devices cleared for scalp use and follow the manufacturer’s protocol.
Agathon