THE SCIENCE

What the evidence actually says about red light therapy.

Red light therapy has a real research base and a serious exaggeration problem, and the two are hard to tell apart from a product page. Every figure below traces to a named primary source, linked at the bottom. Claims we could not trace — including several that circulate widely — were removed rather than softened.

Last reviewed September 2026 · 22 primary sources

Light, not injury.

Red light therapy is the use of specific red and near-infrared wavelengths, delivered by LEDs, to stimulate biological activity in skin and the tissue beneath it.

What separates it from most other skin treatments is that it does not work through damage. Ablative lasers and chemical peels create controlled injury that the skin then repairs. This does neither — it is non-ablative and non-thermal.1,2,3

Also called

Photobiomodulation (PBM), or low-level light therapy. The terms describe the same underlying mechanism, though devices and protocols vary widely.

Non-ablative, non-thermal

No tissue is removed and no controlled wound is created. The energy levels are low compared with lasers used for cutting or coagulating tissue.1

No ultraviolet

These devices emit no UV. Near-infrared above roughly 780nm also falls outside the human visible range — the ICNIRP defines visible light as 380 to 780nm — which is why a near-infrared emitter looks dim or dark in use.4

Where this came from.

The wellness industry did not invent this. The origin is genuinely odd, well documented, and routinely retold with details that do not survive checking — so here it is with the parts we could verify, and a note about the part we could not.

ORIGINS
Four moments that built the field.
1
1967
An accidental observation
Endre Mester, at Semmelweis University in Budapest, found that shaved mice exposed to a low-powered ruby laser regrew hair faster than untreated controls. The work was published in Radiobiologia Radiotherapia in 1968. [5]
2
1990s
NASA plant lighting
LEDs developed by Quantum Devices with the Wisconsin Center for Space Automation and Robotics were used to grow plants aboard the Space Shuttle, including potatoes in 1995. NASA-funded medical research followed at Marshall Space Flight Center, with Dr. Harry Whelan of the Medical College of Wisconsin as clinical lead. [6]
A published anecdote
NASA's own Spinoff publication records that scientists working under the plant lighting noticed abrasions on their hands appeared to heal faster than usual. That is an anecdote NASA published — not a research finding.
3
2003
An FDA clearance
WARP 10, a device from this NASA lineage, was cleared for temporary relief of minor muscle and joint pain. It was not the first LED clearance, and it was not a wound-healing clearance. [6]
4
2025
An evidence-based consensus
A 21-member expert panel published the first evidence-based Delphi consensus on the clinical use of photobiomodulation, in the Journal of the American Academy of Dermatology. [7]
Mester's experiment is usually retold as a study of whether laser light causes cancer. The 1968 paper describes hair regrowth in shaved mice; we could not verify the cancer framing against a primary source, so it is not here.
MECHANISM

The leading hypothesis — and why it is still a hypothesis.

Red light therapy has a proposed biological mechanism, which is more than most of the wellness category can say. What it does not have is a settled one.

You will see the mitochondrial explanation stated as established fact on product pages. The researchers who study it are more careful: reviews describe the mechanism as not fully understood, and one of the field's most-cited authors co-published a 2022 paper titled Mechanisms Beyond Cytochrome c Oxidase.1,8

01

Absorption in the mitochondria

Red and near-infrared light is thought to be absorbed by cytochrome c oxidase, the terminal enzyme in the mitochondrial electron transport chain, increasing ATP production. This is the leading hypothesis, not a demonstrated chain of cause and effect.8

02

Nitric oxide release

A proposed secondary effect. Experimental studies associate it with local vasodilation and increased blood flow. It has not been shown to be the cause of any cosmetic outcome.1

03

Fibroblast response

In laboratory and small clinical studies, red light has been associated with increased type-1 procollagen and reduced MMP-1 — the enzyme that breaks collagen down.9

WAVELENGTH

What the numbers on the box mean — and what they don't.

Penetration depth does increase with wavelength. That ordering is real, and it is the reason serious devices pair a visible red wavelength with an invisible near-infrared one.

What is not reliable is any specific millimetre figure. Monte Carlo modelling of human skin put maximum penetration at about 5.4mm when penetration was defined as 1% of surface intensity — and about 0.37mm when the threshold was set at 13.5%. Same tissue, same model, a fifteen-fold difference in the answer depending on where you draw the line. At 1mm deep, delivered energy is roughly 5% of the surface value for red and near-infrared alike.10

570 nm660 nm850 nm
THE THREE BANDS
What each band has been studied for.
~590 nm
Amber / Yellow
Studied for erythema and pigmentation. The shortest of the three wavelengths, and the shallowest.
633–660 nm
Red
Where the collagen-endpoint evidence sits. Every wrinkle and procollagen trial cited below used this band.
~830 nm
Near-Infrared
Penetrates further and is used for deeper targets. Above roughly 780nm it falls outside the visible spectrum.
These are the bands each wavelength has been studied in — not a map of which skin layer each one treats. The American Academy of Dermatology notes that comparing studies is difficult because researchers use different devices, wavelengths and protocols. No validated wavelength-to-effect map exists. [2]
CLINICAL EVIDENCE

What the trials measured.

This is not a large literature, and it is not a uniform one. What follows are the quantified skin-rejuvenation results we could trace to a primary source and state accurately. Where a number needs a qualifier to be honest, the qualifier is on the card.

MEASURED OUTCOMES
Three findings, stated the way the papers state them.
~30%
Periocular wrinkle volume
RCT · 137 women aged 40–65
Split-face design, 660nm against 590nm, 10 sessions over four weeks. A published response letter disputes the paper's reported dosimetry, so we quote the outcome and not its energy figures.
Mota et al., Photobiomodul Photomed Laser Surg, 2023 [11,12]
up to 36%
Wrinkle reduction, best-performing arm
Split-face RCT · n=76 · four arms including sham
Twice weekly for four weeks. 36% is the maximum recorded in the strongest arm — not an average across participants, and not what every arm achieved.
Lee et al., J Photochem Photobiol B, 2007 [13]
31%
Type-1 procollagen — in the lab
Laboratory-grown human skin tissue · 660nm
MMP-1, the enzyme that degrades collagen, fell 18% in the same experiment. This is engineered tissue with a correlated clinical arm — not a measurement of collagen density in living skin.
Barolet et al., J Invest Dermatol, 2009 [9]
Three studies is not a large body of evidence. This is what these papers measured, in the populations they measured it in.

The broadest look at the field is a 2023 systematic review and meta-analysis that screened 554 articles and included 31. It found statistically significant pooled effects for red and blue LED in acne. That is a result about acne specifically — the same review noted inconsistency between studies on wound healing and atopic dermatitis outcomes.14

One 2023 study is worth singling out because it tested a mask rather than a clinical panel: a 630nm LED mask used twice weekly for three months documented reversal of photoaging signs. The device studied was a commercial product developed with its manufacturer, which is worth knowing when weighing the result.15 For a broader clinical overview, the Journal of the American Academy of Dermatology published a CME review of photobiomodulation's dermatological applications in 2024.16

What a dose in the literature actually looks like.

Device marketing tends to lead with session length, because three minutes sounds better than twenty. Session length on its own means nothing without the power behind it. These are the parameters the published trials actually used.

DOSE
The numbers the trials used.
ParameterWhat the literature reportsWhy it matters
Irradiance (mW/cm²)Most dermatologic studies fall around 20–60. The broader published range spans roughly 10–100.Power delivered per square centimetre of skin. This is the number that determines whether a session delivers a meaningful dose.
Fluence (J/cm²)Commonly 4–18 per session, though some published trials used 66–126.Total energy per square centimetre. There is no single agreed therapeutic figure — the spread is the point.
FrequencyTwo to three sessions per week. The 2007 split-face trial and the 2023 mask study were both twice weekly.Higher-frequency schedules quoted on product pages are not what the trials tested.
Time to endpointEight to twelve weeks is where trials measure meaningful change in wrinkles and firmness.Surface changes tend to be noticed before structural ones — a rule of thumb, not a cited finding.
LED count and coverageEven distribution across the treatment area.Gaps mean untreated skin. A raw LED count tells you nothing about whether coverage is uniform.
Irradiance and fluence reporting across this literature is inconsistent and sometimes internally contradictory; a 2025 paper in PLOS One documents methodological and reporting problems in LED therapy research. Treat every figure here as the range the field works in, not a specification. [17]
REGULATION

Cleared is not approved.

“FDA-cleared” and “FDA-approved” are not synonyms, and the difference is not cosmetic. LED masks are regulated as Class II devices under 21 CFR 878.4810 and reach the market through 510(k) clearance.18

A 510(k) submission demonstrates that a device is “substantially equivalent” to one already legally on the market. Premarket approval — what “FDA-approved” actually describes — requires, in the FDA's words, “valid scientific evidence that provides reasonable assurance that the device is safe and effective.” Consumer LED masks do not take that pathway.19

TWO DIFFERENT STANDARDS
What the FDA actually did.
510(k) clearance — the pathway LED masks use
  • Class II device under 21 CFR 878.4810.
  • The manufacturer demonstrates 'substantial equivalence' to a device already on the market.
  • No requirement to run a new trial proving this particular device works.
  • The American Academy of Dermatology notes clearance means a device is considered safe — not necessarily effective. [2]
Premarket approval — what 'FDA-approved' means
  • Reserved for higher-risk device classes.
  • Requires 'valid scientific evidence that provides reasonable assurance that the device is safe and effective.'
  • A materially more demanding standard.
  • Not the pathway consumer LED masks take. [18]
CLEARED INDICATIONS
What these devices are cleared to do.
IndicationProduct codeSetting
Full-face wrinkle reductionOHSOver-the-counter
Mild-to-moderate acneOLPOver-the-counter
Temporary relief of minor muscle and joint pain; temporary increase in local blood circulationNHN—
Promoting hair growth in men with androgenetic alopecia (laser combs)OAP—
Wound healing—Not a cleared consumer indication
Product codes are FDA classification identifiers. [20] The research narrative starts with wound healing, and it is not on the consumer list. The FDA has issued no consumer safety communication about LED masks, but it has sent warning letters to companies making claims beyond their clearance — a 2019 letter stated the agency was unaware of any low-level LED device cleared to treat inflammation or neuropathy. [21] The agency also published draft guidance in January 2023 covering 510(k) submissions for photobiomodulation devices. [22]
THE LIMITS

What the evidence still doesn't settle.

This page exists because the category is sold with more confidence than the literature supports. The honest version is more useful than either the hype or the dismissal — and that means being clear about where the research stops.

01

Dosimetry reporting is inconsistent

Trials report power and energy in ways that cannot be compared across studies, and sometimes contradict themselves internally.17

02

Many rejuvenation studies are small, uncontrolled, or industry-linked

Including one cited on this page. Sample sizes in the dozens are normal here, and self-reported outcomes are common.

03

The 2025 consensus settles several indications — not this one

The expert panel found photobiomodulation effective for peripheral neuropathy, androgenetic alopecia, wound ulcers of several etiologies, decubitus ulcers, pain from diabetic foot ulcers, and acute radiation dermatitis. Skin rejuvenation is not on that list; the panel called for ongoing research into aesthetic applications.7

04

Clearance is not proof of effectiveness

A 510(k) establishes that a device is substantially equivalent to something already being sold. It is not a finding that the device works.18

REFERENCES

Every source, linked.

Each number in the text above links to its entry here. Where a claim could not be traced to a primary source, it is not on this page.

  1. 01

    Avci P, Gupta A, Sadasivam M, et al. Low-level laser (light) therapy (LLLT) in skin: stimulating, healing, restoring. Seminars in Cutaneous Medicine and Surgery. 2013;32(1):41–52.

    PubMed Central
  2. 02

    American Academy of Dermatology. Red light therapy: Effective for skin conditions?

    aad.org
  3. 03

    Cleveland Clinic. Red Light Therapy: Benefits, Side Effects and Uses.

    Cleveland Clinic
  4. 04

    International Commission on Non-Ionizing Radiation Protection. Visible radiation (380–780 nm).

    ICNIRP
  5. 05

    Mester E, Szende B, Gärtner P. The effect of laser beams on the growth of hair in mice. Radiobiologia Radiotherapia. 1968;9(5):621–626. PMID 5732466.

    PubMed
  6. 06

    NASA Spinoff. NASA Research Illuminates Medical Uses of Light.

    NASA.gov
  7. 07

    Maghfour J, Mineroff J, Ozog DM, et al. Evidence-based consensus on the clinical application of photobiomodulation. Journal of the American Academy of Dermatology. 2025;93(2):429–443.

    JAAD
  8. 08

    Hamblin MR. Mechanisms and Mitochondrial Redox Signaling in Photobiomodulation. Photochemistry and Photobiology. 2018;94(2):199–212. PMID 29164625.

    PubMed
  9. 09

    Barolet D, Roberge CJ, Auger FA, Boucher A, Germain L. Regulation of skin collagen metabolism in vitro using a pulsed 660 nm LED light source: clinical correlation with a single-blinded study. Journal of Investigative Dermatology. 2009;129(12):2751–2759. PMID 19587693.

    PubMed
  10. 10

    Ash C, Dubec M, Donne K, Bashford T. Effect of wavelength and beam width on penetration in light-tissue interaction using computational methods. Lasers in Medical Science. 2017;32(8):1909–1918.

    PubMed Central
  11. 11

    Mota LR, Duarte IDS, Galache TR, et al. Photobiomodulation Reduces Periocular Wrinkle Volume by 30%: A Randomized Controlled Trial. Photobiomodulation, Photomedicine, and Laser Surgery. 2023;41(2):48–56. PMID 36780572.

    PubMed
  12. 12

    Jenkins PA. Response to: Photobiomodulation Reduces Periocular Wrinkle Volume by 30%. Photobiomodulation, Photomedicine, and Laser Surgery. 2023. PMID 37252792.

    PubMed
  13. 13

    Lee SY, Park K, Choi J, et al. A prospective, randomized, placebo-controlled, double-blinded, and split-face clinical study on LED phototherapy for skin rejuvenation. Journal of Photochemistry and Photobiology B. 2007;88(1):51–67. PMID 17566756.

    PubMed
  14. 14

    Ngoc LTN, Moon JY, Lee YC. Utilization of light-emitting diodes for skin therapy: Systematic review and meta-analysis. Photodermatology, Photoimmunology & Photomedicine. 2023;39(4):303–317.

    Wiley
  15. 15

    Couturaud V, Le Fur M, Pelletier M, Granotier F. Reverse skin aging signs by red light photobiomodulation. Skin Research and Technology. 2023;29(7):e13391.

    Wiley
  16. 16

    Mineroff J, Maghfour J, Ozog DM, Lim HW, Kohli I, Jagdeo J. Photobiomodulation CME part II: Clinical applications in dermatology. Journal of the American Academy of Dermatology. 2024;91(5):805–815.

    JAAD
  17. 17

    Methodological issues in visible LED therapy dermatological research and reporting. PLOS One. 2025.

    PLOS One
  18. 18

    U.S. Food and Drug Administration. Device Approvals, Denials and Clearances.

    FDA.gov
  19. 19

    U.S. Food and Drug Administration. Consumer Update: Is It Really 'FDA Approved'?

    FDA.gov
  20. 20

    U.S. Food and Drug Administration, Product Classification Database. Product codes OHS, OLP, NHN and OAP.

    FDA.gov
  21. 21

    U.S. Food and Drug Administration. Warning Letter: Vevazz LLC, December 26, 2019.

    FDA.gov
  22. 22

    U.S. Food and Drug Administration. Photobiomodulation (PBM) Devices — Premarket Notification [510(k)] Submissions. Draft guidance, January 2023.

    FDA.gov