Tip · Light

Red light therapy

Q: Does a red light panel really do anything?

For some narrowly defined endpoints, yes: hair density in androgenetic alopecia with eSMD 1.32, fibromyalgia fatigue with 1.25, knee osteoarthritis function with 0.65, cognitive function with 0.49 and burning mouth pain with −0.92, all with moderate certainty of evidence. For more energy or general recovery in healthy people, this level of support does not exist.

Q: Which wavelength is the right one?

The body of studies gives no answer to that. The wavelengths in the muscle soreness literature alone range from 660 to 950 nm. The umbrella review explicitly names the lack of standardization of wavelength, dose and irradiance as the reason why the results do not fit together.

Q: How often and for how long should you use a panel?

In the double-blind skin study, 21 minutes per session twice a week over four weeks was just as effective as three times a week. For other applications there are no comparably tested regimens, and for full-body panels none at all.

Q: Use it before or after training?

According to the current meta-analysis, before training. Photobiomodulation before exercise significantly reduced muscle soreness by MD −12.27 points, while the methods applied after exercise showed no effect in the same paper. The certainty of evidence is low.

Q: Is red light dangerous for the eyes?

For devices that deliberately shine into the eye, retinal damage has been documented in case reports, and in a three-year study one child developed reversible retinal hyperreflectivity on OCT. For home panels this means: do not look into them. Systematic data on the eye safety of full-body panels in adults are not available.

Q: Does red light also work on wounds and scars?

The umbrella review covers 15 clinical conditions; wound healing is not among the five endpoints with moderate certainty of evidence. There is literature on it, but none that could be described as reliably quantified.

Red and infrared light: many small human studies with very different devices and doses – they can hardly be compared.

Red and infrared light is one of the topics where the number of human studies and the usefulness of those studies are furthest apart. There are dozens of them, but with such different devices, wavelengths and doses that they can hardly be compared. This page therefore answers only one question: what can a healthy person expect from a home panel they have bought?

In short

For some narrowly defined, mostly disease-related endpoints there are solid effects: hair density in androgenetic alopecia with eSMD 1.32, fibromyalgia fatigue with 1.25, knee osteoarthritis function with 0.65, cognitive function with 0.49. For muscle soreness, use before exercise lowers the score by MD −12.27 points, with low certainty of evidence. This was measured, however, with locally or transcranially applied protocols, not with full-body panels in the living room. Of 35 endpoints, five reached moderate certainty of evidence, none high. And because there is no standardized home dose, no study result can be cleanly reproduced.

What is behind it

The broadest synthesis is an umbrella review of meta-analyses of randomized trials: 15 meta-analyses, 204 RCTs, more than 9,000 participants, 35 endpoints in 15 clinical conditions. Significant effects were found for 12 endpoints, but moderate certainty of evidence for only five of them.

These five are: pain in burning mouth syndrome (eSMD −0.92; 95% CI −1.38 to −0.46), functional limitation in knee osteoarthritis (0.65; 0.14 to 1.15), fatigue in fibromyalgia (1.25; 0.63 to 1.87), hair density in androgenetic alopecia (1.32; 1.00 to 1.63) and cognitive function (0.49; 0.14 to 0.84). All others were at low or very low certainty, mostly because of heterogeneity or small-study effects. P-curve and funnel plot analyses found indications of publication bias for several endpoints.

How it is supposed to work

The common explanation: red and near-infrared light is absorbed by cytochrome c oxidase, complex IV of the mitochondrial respiratory chain. As a result, electron transport and ATP production increase, nitric oxide is released, and low-dose oxidative stress triggers hormetic signaling pathways. The review literature explicitly lists this chain as “proposed mitochondria-centered mechanisms of action”.

Crucial for the assessment: in humans it has not been quantified. There is no human study showing how much light of a particular wavelength produces how much additional ATP at what tissue depth. The hypothesis is plausible; the calculation for it is missing.

What a panel actually delivers of this

The endpoints with the best support are all disease-related and applied locally or transcranially: oral mucosal pain, knee osteoarthritis function, fibromyalgia fatigue, hair density and cognitive function. “Cellular energy in healthy people” is not a studied endpoint. For sport there is a single finding with numbers, and it is tied to a condition that rarely appears in advertising: it occurs with use before exercise, not after.

What is well supported

What holds up is the breadth of human evidence for local, clinical applications. Five endpoints reach moderate certainty of evidence, and the effect sizes there are not small: eSMD 1.32 for hair density in androgenetic alopecia, 1.25 for fibromyalgia fatigue, 0.65 for function in knee osteoarthritis, 0.49 for cognitive function and −0.92 for pain in burning mouth syndrome. Added to this is the sports finding: in 19 studies with n = 672 healthy adults, photobiomodulation before exercise reduced muscle soreness by MD −12.27 points (95% CI −18.14 to −6.40; I² = 48%). And in the only double-blind study on skin, subjective satisfaction at 79.6% and 73.4% respectively was significantly higher than under sham treatment.

What the studies show

The broadest synthesis (Son et al. 2025)

Umbrella review of meta-analyses of randomized trials, PROSPERO CRD42023495502, search up to December 8, 2023, assessment with AMSTAR 2 and modified GRADE. Scope: 15 meta-analyses, 204 RCTs, more than 9,000 participants, 35 endpoints in 15 clinical conditions. Result: significant effects for 12 of 35 endpoints, moderate certainty of evidence for five, high for none. The authors’ conclusion: before broad clinical use, higher-quality studies and standardization of protocols are needed.

Muscle soreness before or after training (Canez et al. 2025)

Systematic review with meta-analysis, only RCTs with healthy, symptom-free adults aged 18 and over. 19 studies, n = 672. Only photobiomodulation applied before exercise significantly reduced muscle soreness (MD −12.27 points). Electrical stimulation after exercise showed no effect (MD −4.60), and neither did pneumatic compression (MD −2.20).

Facial wrinkles under double-blind conditions (Bragato et al. 2025)

Randomized, sham-controlled, double-blind, n = 95 women aged between 45 and 60. Red LED mask, 660 ± 10 nm, 6.4 mW/cm², 8.05 J/cm², 5.02 mW, 21 min per session, three or two times a week over four weeks. On the Wrinkle Assessment Scale, the rating by three blinded experts, there was no significant difference. In the ImageJ measurement of the same photos, however, there was (p < 0.001).

Eye safety with red light into the eye (Wu et al. 2026; Nie et al. 2026)

For low-level red light therapy for myopia control in children, case reports of retinal damage and transient visual disturbances have been documented. In a three-year comparative study with 86 children, 40 of them with red light, one child showed reversible retinal hyperreflectivity on OCT without symptoms.

What the studies do not show

The core problem is the lack of comparability, and the umbrella review itself names it as the reason for the heterogeneity. In the muscle soreness literature alone, the wavelengths used range from 660 to 950 nm; one of the reviews on this included 14 studies, of which only four provided usable data for the quantitative synthesis. Anyone who buys a panel is not buying a standardized procedure. In addition, more than half of the endpoints have low or very low certainty, and several show indications of publication bias.

For skin wrinkles, the result depends on the measurement method. In the only sham-controlled double-blind study with n = 95, the effect was not detectable in the clinical expert rating, but it was in the software measurement of the same photos. That is exactly the pattern typical of surrogate markers. And the full-body and longevity endpoints advertised for panels have simply not been studied: the five endpoints with moderate certainty of evidence are oral mucosal pain, knee osteoarthritis function, fibromyalgia fatigue, hair density and cognitive function.

How to do it

What was actually done in the studies can only be stated for two applications. For skin: red LED mask, 660 ± 10 nm, 6.4 mW/cm², 8.05 J/cm², 5.02 mW, 21 minutes per session, two or three times a week over four weeks — with two sessions a week sufficing for the same result as three. For muscle soreness: application before exercise, with wavelengths between 660 and 950 nm, applied at one to six points on the affected muscle.

For everything else there is no reliable template. A standardized specification of distance and irradiance for home full-body panels does not exist; the widespread figure of a 30–60 cm distance cannot be supported, nor can daily use for 10–20 minutes.

Safety

In the evidence reviewed, photobiomodulation is considered non-invasive and well tolerated; the umbrella review reports no systematic safety signals. Three points nonetheless deserve attention. The eyes are the most critical area: for devices that deliberately shine into the eye, retinal damage has been documented in case reports. For home panels this means: do not look into the light source, use eye protection. Second, dose specifications on consumer devices are not standardized, and even in the specialist literature dosimetry information is regularly incomplete or inconsistent. Third, the endpoints with the best data are clinical conditions — burning mouth syndrome, knee osteoarthritis, fibromyalgia, androgenetic alopecia. These are cases for medical assessment, not for self-treatment.

BK-Score Thin human evidence

Human evidence		4
Mechanism		5
Safety data		7
Hype gap		3
Track record of use		6

There are many small human studies on skin, muscle recovery and wound healing, but with very different devices, wavelengths and doses – they can hardly be compared. The mechanism via cytochrome c oxidase is plausible, but not quantified in humans.

The score rates the state of knowledge, not the effect. “Safety data 9” means well studied – not harmless.
Subjective assessment by Biohacking Kompakt based on published scoring rules – not a scientific rating and not a medical recommendation. Rules and all ratings (German)

Frequently asked questions about red light therapy

Does a red light panel really do anything?