Red Light Therapy for Scars: What the Seven Controlled Trials Found

Affiliate disclosure: This post contains affiliate links. See our affiliate disclosure for details.

Scars are the rare red light topic where the answer is not "there is no evidence." There is evidence. There are randomised, placebo-controlled human trials, several of them split-scar designs where one half of a person's own scar is treated and the other half is not — which is about the strongest design you can run on a problem like this. A scoping review published in Lasers in Medical Science in 2026 collected them: seven studies, 297 patients.

So we did the thing we would want a review site to do, and read all seven primary papers alongside the review that collected them. The direction of effect is consistent and mostly favourable. The size of it, in the original papers, is smaller than the summary implies — and in two cases the numbers in the review's tables are not the numbers in the papers.

We own one red light panel and use it most days, mostly on a shoulder. We have not treated a surgical scar with it, so nothing below is a personal before-and-after. What follows is the published record, the arithmetic on what these devices actually emit, and the part almost nobody writes about: the dose you would need to run is far smaller than the dose you are probably running.

Last reviewed: September 2026

---

Quick Comparison: At-Home LED Devices for Scar Work

Device	Best for	Price	Wavelengths	Irradiance (manufacturer)	One session at the maker's own timing	Rating
Hooga HG300	Body scars, non-contact	~$199	660 + 850nm	Over 73 mW/cm² at 6 in	~44 J/cm² at 10 min	4.3
Omnilux Contour Face	Facial surgical scars	~$395	633 + 830nm	~30 mW/cm²	18 J/cm² at 10 min	4.4
Bestqool BQ40	Small, fixed targets	~$149	660 + 850nm	70 mW/cm² at 3 in	126 J/cm² at 30 min (maker's figure)	4.0
Hooga Red Light Therapy Belt	Wrapping a limb or torso scar	~$249	660 + 850nm	Over 70 mW/cm² at the surface	~4 J/cm² at its 1-minute setting	4.1
FlexBeam	Targeted, strapped-on sessions	~$599	610–650nm (peak 622) + 800–850nm (peak 814)	110 mW/cm²	66 J/cm² per 10-min program	4.0
Celluma HOME Series 2	Contouring over an awkward site	~$795	465 + 640 + 880nm	Not published	Not calculable — no irradiance figure	3.9

Prices and specs checked 5 September 2026 against each brand's own US storefront, and prices move — confirm on the brand's site before you buy. FlexBeam's standing price is $599; Recharge Health was running 20% off, taking it to $479, on the day we checked. The dose column is our arithmetic, not a manufacturer claim: irradiance multiplied by the session length the brand itself recommends. Treat it as order of magnitude. Brands measure irradiance in ways that flatter them, solar-meter readings routinely come in lower than published figures, and the number changes the moment you move the panel.

---

The seven trials, and what happens when you read them

The review is Gaumond, Dreifus, Mittal and Eber, "Red and near-infrared photobiomodulation for burn, hypertrophic, and post-surgical scars: a scoping review of clinical trials," Lasers in Medical Science 2026;41:80 (doi:10.1007/s10103-026-04875-8, PMID 42026334). It searched PubMed, Embase and MEDLINE through October 2025, screened 118 records, and included seven.

Its conclusion is measured and we have no quarrel with it: red and near-infrared photobiomodulation "represents a safe, non-invasive adjunct for scar management," with red suiting early superficial scars and near-infrared suiting deeper or more mature ones, and larger controlled trials still needed. The review says plainly that light should be an adjunct rather than a replacement for silicone sheeting, corticosteroid injection, pressure therapy or laser resurfacing.

The trouble is what happens between the primary papers and the summary tables.

Four of the seven used lasers, not LEDs

The review's inclusion criteria cover LED or low-level laser therapy, so including laser studies is entirely legitimate. What is harder to defend is that its results tables label them as LED.

• Alsharnoubi et al. 2018 (Lasers Med Sci 33:1991–1995) is listed as 633nm LED. The paper treated 15 children with post-burn hypertrophic scars using a helium-neon laser, which emits at 632.8nm.
• Gaida et al. 2004 (Burns 30:362–367) is listed as 670nm LED. The paper used a 400 mW 670nm Softlaser.
• Carvalho et al. 2010 (Photomed Laser Surg 28:417–422) is listed as 830nm LED. The paper used an 830nm GaAlAs diode laser at 40 mW through a 0.08 cm² aperture, 26 seconds and 1.04 J per point, ten points per scar.
• Freitas et al. 2013 (J Cosmet Laser Ther 15:171–176) is listed as an 808nm LED device. The paper is titled "Efficacy of low-level laser therapy on scar tissue" and blinded its placebo group by switching the laser off.

This matters for a reader holding a $199 panel. A laser delivers coherent light through a small aperture at a defined spot size, which is how Carvalho's team could state an energy density of 13 J/cm² per point and mean it. An LED array sprays incoherent light over a wide area from whatever distance you happen to hold it. The biology may well be the same. The dosimetry is not, and the four trials with the tidiest numbers are the four that were not done with the technology being sold to you.

That leaves three LED trials. Here is what is in them.

The largest LED trial did not measure a scar

Ye and Xiang 2025 (Aesthetic Plast Surg 49:655–663) is 145 of the review's 297 patients — half the total on its own. It randomised blepharoplasty patients to 830nm LED phototherapy or standard post-operative care and reported significantly better swelling scores, pain, wound-healing grading and anxiety scores.

Those are post-operative complication measures. They are not scar measures. No Vancouver Scar Scale, no POSAS, no imaging of the healed scar appears in the abstract. It is a perfectly good trial of post-operative recovery, and it is being counted towards a total about scarring.

Take it out and the trials that actually measured a scar cover 152 patients, not 297.

The best-designed LED trial missed its primary endpoint

Kurtti et al. 2021 (J Biophotonics 14:e202100073) is the one built to consumer-relevant standards: a randomised, mock-controlled, single-blind, dose-ranging, split-face phase II trial in 30 patients after facial surgery, starting a week after surgery, at cumulative fluences of 160, 320 or 480 J/cm² three times weekly for three weeks.

Its primary endpoint was the difference in scar pliability between treated and control sites. The paper's own abstract: "There were no significant differences in scar pliability between treated and control scars."

The interesting result is a secondary one, and it is genuinely interesting. The moderate-fluence arm showed a 77.8% reduction in induration against 50% on the untreated side, while the highest fluence arm did worse than its control — 50.0% against 71.4%. More light was not better light. Two patients blistered and one had localised swelling at the treatment site — the only blistering reported across the seven trials — while transient erythema and warmth were universal.

The review reports the percentages prominently and does record "no group-wise significance" in its table. A reader skimming the narrative would come away with the opposite impression of a trial that missed its primary endpoint.

That leaves one home-use trial

Kim et al. 2022 (Lasers Med Sci 37:3583–3590, PMID 36045183) is the study to read if you are considering buying something. It is randomised, double-blind and sham-device-controlled: 43 thyroidectomy patients, 21 treated and 22 shams, using a home-based 830nm LED device daily from one week after surgery for four weeks, at 4.5 J/cm² per session.

At six months the treated group had significantly higher satisfaction (p = 0.008) and global assessment scores (p = 0.002), lower Vancouver Scar Scale scores (p = 0.004), and lower reported pain. No adverse events.

And then the part that gets left out. The trial also assessed the scars with a three-dimensional skin imaging device, measuring colour, height, pigmentation and vascularity. All four favoured the treated group. None of them reached statistical significance.

So the honest summary of the whole at-home LED case for scars is one 43-patient trial in which the scales people filled in improved and the machine that photographed the scar could not confirm it. That is not nothing — patient-reported outcomes and the Vancouver Scar Scale are the standard currency of scar research, and a double-blind sham-controlled design is hard to argue with. It is also a long way from the certainty of the marketing.

Two numbers we could not reconcile

Two of the review's summary figures do not match the papers they cite. We are flagging the discrepancy rather than resolving it — we read the published abstracts, not the full texts, and a full text can contain subgroup numbers that an abstract does not.

Gaida 2004. The review's table gives Vancouver Scar Scale scores falling from 7.10 to 1.42 in treated areas against 5.86 to 5.40 in placebo areas. The paper's abstract gives 7.10 ± 2.13 to 4.68 ± 2.05 treated, and 6.10 ± 2.86 to 5.88 ± 2.72 control. The review's own narrative, two sentences later, reports pain falling "from 3.81 ± 3.07 to 1.42 ± 1.67" — the same 1.42 ± 1.67 it places in the scar-score row. The difference is not cosmetic: 7.10 to 4.68 is a 34% improvement, 7.10 to 1.42 is an 80% one.

Freitas 2013. The review describes a split-scar design with one scar randomised to active light and another to placebo within each patient, and reports significant improvements in colour (p = 0.004) and elasticity (p = 0.016). The paper describes 17 volunteers randomly assigned to two separate groups, nine active and eight placebo, and its abstract states: "After 5 weeks, there were no statistically significant differences in any variable between both the groups." The significant result it does report, macroscopic appearance at p = 0.003, is a within-group change in the treated arm — which is exactly the comparison a placebo arm exists to distrust.

None of this makes photobiomodulation useless for scars. It makes the field smaller and softer than the summary suggests, and it is the reason we are not going to tell you a panel will flatten your scar.

---

The dose problem, which is arithmetic rather than opinion

Here are the per-session doses in the trials that reported them:

Trial	Per session	Sessions	Cumulative
Gaida 2004 (burn scars)	4 J/cm²	16	64 J/cm²
Freitas 2013 (hypertrophic)	4 J/cm²	15	60 J/cm²
Kim 2022 (home use, thyroidectomy)	4.5 J/cm²	~28	126 J/cm²
Carvalho 2010 (hernia repair)	13 J/cm²	4	52 J/cm²
Kurtti 2021 (facelift, best arm)	~36 J/cm²	9	320 J/cm²

Now run the same arithmetic on the hardware. Bestqool publishes, in its own spec block for the BQ40, a dose of 126 J/cm² after 30 minutes at three inches. That is the entire four-week cumulative dose of the only home-use trial, delivered in one sitting.

Hooga publishes over 73 mW/cm² at six inches for the HG300. At that figure, Kim's 4.5 J/cm² session takes about 62 seconds. Omnilux's Contour Face runs a ten-minute program at roughly 30 mW/cm², which works out at 18 J/cm² — four times the trial dose, and the lowest per-session figure of any device here, because the session length is not yours to extend.

The point is not that a long session will hurt you. It is that photobiomodulation has a well-documented biphasic dose response — too little does nothing, too much stops helping — and Kurtti's highest-fluence arm did worse than its own control while producing the only blisters in the whole literature. Every incentive in the consumer market pushes the other way: bigger panels, higher irradiance, longer sessions, a number on the box that beats the competitor's number.

If you want to reproduce what the trials did, you want the small dose, several times a week, for weeks. That is an unusual thing for a review site to tell you, and it makes the cheap panel the better buy rather than the worse one.

---

What the guidelines actually recommend, and where light sits in them

Any honest page on this has to say what the first-line treatment is, and it is not light.

The international practical guidelines on scar management — Monstrey and colleagues, J Plast Reconstr Aesthet Surg 2014, drawn up by 24 experts across specialities — state that silicone sheeting or gel is universally considered the first-line prophylactic and treatment option for hypertrophic scars and keloids. Alongside it sit the unglamorous basics applied before, during and after surgery: reducing skin tension, taping, hydration and ultraviolet protection of the early scar.

For redness and vascularity, the modality with the clinical track record is the pulsed dye laser at 585nm, delivered in a clinic. For thickened keloids, intralesional corticosteroid injection. For texture, fractional ablative or non-ablative resurfacing. The 2026 scoping review places photobiomodulation explicitly alongside these as an adjunct, not instead of them.

So the sensible position for a home device is: silicone and sun protection first, a clinician's opinion on anything raised, thickened or spreading, and light as the cheap, low-risk thing you add on top. Anyone with a keloid, a scar that is growing, or a wound that has not fully closed should be talking to a dermatologist rather than shopping.

---

Timing matters more than the device

The clearest pattern across the seven trials is about when, not what.

Early scars respond to red. In the burn work, scars younger than 12 months improved more than older ones. Kurtti started treatment one week after surgery; Kim started one week after surgery; Carvalho started 24 hours after surgery. The window when a scar is actively remodelling, roughly the first year, is when the trials intervened.

Mature scars are the near-infrared case. Freitas found the greatest improvement in scars older than 24 months, and reported a trend towards reduced thickness in older scars specifically — which is the one encouraging thing in a trial whose between-group comparison came out null. The review's rationale is penetration depth: 633 to 670nm works on superficial dermal layers, 808 to 833nm reaches deeper fibroblast and vascular networks.

This is also where the consumer wavelength mismatch shows up. The scar trials clustered at 808 to 833nm. Most panels sold as "near-infrared" run 850nm, just outside that band, and the Celluma HOME runs 880nm, well outside it. The masks are the closest match: Omnilux Contour at 633 and 830nm sits exactly on the trial wavelengths, and FlexBeam's near-infrared peaks at 814nm. Whether 20nm changes anything is not established, and we are not going to pretend it is — but if you are choosing between two devices at a similar price, the one that matches the studied band is the better-reasoned pick. Our red light therapy wavelengths guide goes through what each band is actually documented to do.

---

Acne scars and stretch marks are a different problem

The most common search behind this topic is acne scarring, and it needs separating out.

An atrophic acne scar — the ice-pick, boxcar and rolling depressions — is a volume deficit. Tissue is missing. Searching the literature for a photobiomodulation or LED trial in atrophic acne scars returns nothing usable; on 5 September 2026 we ran three query shapes across Europe PMC pairing acne or atrophic scars with photobiomodulation, LED and low-level laser, and every clinical result that came back was a fractional laser, picosecond laser or microneedling study. That is not proof no such trial exists — a badly-shaped search can miss real papers, and we have been caught by exactly that before — but it is the state of what is indexed and findable.

Omnilux is unusually straight about this on its own FAQ page, writing of its Clear device that it "may help to reduce the chances of scarring and may soften the appearance of acne scarring" but "will not soften the appearance of atrophic scars." That is a brand declining to claim something it could easily have implied, and it is the correct answer.

What red light plausibly does for acne is upstream: fewer and less inflamed lesions mean less post-inflammatory damage to scar over. The redness left behind after a spot clears — post-inflammatory erythema, not a scar at all — fades on its own and responds to vascular lasers. For the depressions, the treatments with evidence are microneedling and fractional resurfacing, and they are clinic procedures.

Stretch marks are the same shape of problem in a different tissue, and the studies that exist pair light with something else, usually an ablative laser, so the light's independent contribution cannot be read off them.

---

How to choose, if you are buying one anyway

Match the site, not the marketing. A scar is a small target. A 300-watt panel that covers your whole torso is not more useful than a small one for a six-inch incision; it is just harder to hold at a consistent distance. For a facial surgical scar, a mask solves the distance problem entirely, because it sits at a fixed geometry every time. For a limb or a C-section scar, a wrap does the same job.

Prefer a device that constrains the dose. This is the reverse of the usual advice. The Omnilux Contour's ten-minute program and the Hooga belt's one-minute timer floor are features here, not limitations. A panel with no timer granularity invites you to stand in front of it for twenty minutes because that feels more thorough.

Non-contact for anything recent. Pressure, occlusion and sweat on a healing incision are all reasons the trials used devices held off the skin. Masks and wraps touch. Panels do not. Once a scar is mature the distinction matters much less.

Check what the clearance is actually for. None of these devices is FDA-cleared for scars. Omnilux Contour is cleared for fine lines and wrinkles. Celluma HOME is cleared for aging skin and pain relief. Recharge Health states FlexBeam is FDA cleared but does not name the indication on its product page. Bestqool's site lists FDA among its certifications without stating a cleared indication, which is not the same thing as a clearance for a device. Hooga makes no clearance claim on the HG300 page at all, which is the most honest of the set.

Do not treat an open wound. Every trial here started on a closed incision — a week post-op in the surgical studies, 24 hours in Carvalho's, where the wound was sutured. If your wound has not closed, or is red, hot or weeping, that is a doctor's problem and not a panel's. The same goes for anything on or near the eye, and for anyone on a photosensitising medication such as isotretinoin. Our rundown of red light therapy side effects covers the contraindications in more detail, and the red light therapy for skin guide covers what the dermatological evidence supports more broadly.

If you want to see how these panels behave in ordinary use before deciding, our Hooga red light review goes through the build and the real-world irradiance question, and the red light therapy hub collects the rest. More about how we assess this gear, and what we do and do not own, is on our about page.

---

FAQ

Does red light therapy actually work on scars?

The evidence is real but thinner than it looks. A 2026 scoping review found seven controlled or prospective trials covering 297 patients, and reported improvements in scar score, pigmentation, thickness, elasticity, pain and itch. Reading the primary papers changes the picture: four of the seven used lasers rather than LEDs, the largest LED trial measured post-operative swelling rather than scar quality, and the best-designed LED trial missed its primary endpoint of scar pliability. The strongest at-home evidence is a single 43-patient double-blind trial of a home-use 830nm device, which improved patient satisfaction and Vancouver Scar Scale scores at six months but showed no statistically significant difference on 3D imaging of colour, height, pigmentation or vascularity. Research suggests a modest benefit as an adjunct. It does not support light as a scar removal treatment.

How long does red light therapy take to work on a scar?

Every trial that found an effect ran for weeks and assessed at months. Kim's home-use trial treated daily for four weeks starting a week after surgery, and reported its results at the six-month follow-up. Gaida treated twice weekly for eight weeks. Freitas ran three sessions a week for five weeks. Carvalho used only four sessions, on post-operative days 1, 3, 5 and 7, and re-assessed at six months. Nothing in the literature suggests a visible change in days, and scar remodelling continues for a year or more regardless of what you do to it, which is exactly why uncontrolled before-and-after photos are so unreliable here.

Can red light therapy help an old scar?

Possibly, and near-infrared is the wavelength with the argument behind it. Freitas found the greatest improvements in scars older than 24 months and a trend towards reduced thickness in older scars, though that trial's between-group comparison was not statistically significant. The 2026 review's framework puts red at 633 to 670nm for early superficial scars, where remodelling is still active, and near-infrared at 808 to 833nm for thicker or more mature ones, on the grounds of deeper dermal penetration. An old, flat, pale scar is unlikely to change much whatever you do. A raised or thickened one is a reason to see a dermatologist rather than to buy a panel.

Does red light therapy help acne scars?

Not the indented kind. Atrophic acne scars — ice-pick, boxcar and rolling — are missing tissue, and searching the clinical literature for an LED or photobiomodulation trial in them returns fractional laser, picosecond laser and microneedling studies instead. Omnilux states on its own FAQ that its Clear device "will not soften the appearance of atrophic scars," which is the honest answer from a brand that could have implied otherwise. Where red and blue light do have acne evidence is upstream, in reducing lesions in the first place, and less inflammation means less scarring to deal with later. The flat red or brown marks left after a spot heals are post-inflammatory pigment change rather than scars, and those usually fade on their own.

Is red light therapy safe to use on a new surgical scar?

The trials suggest yes on a closed incision, with clinician awareness. Kim and Kurtti both began one week after surgery, and Carvalho began 24 hours after a sutured hernia repair. Across the whole seven-trial set the reported adverse events were transient redness and warmth, with two blisters and one episode of swelling, all in the highest-fluence arm of Kurtti's dose-ranging trial. What no trial did was treat an open, unhealed or infected wound, and neither should you. Anyone with a keloid history, a scar that is actively growing, or a wound that has not closed should ask their surgeon before pointing anything at it.

What wavelength is best for scars?

The trials cluster in two bands: red at 633 to 670nm, and near-infrared at 808 to 833nm. The 2026 review's practical framework is red for early, superficial and actively remodelling scars, and near-infrared for thicker, deeper or more mature ones. That creates an awkward mismatch with the consumer market, where most "near-infrared" devices run 850nm, just outside the studied band, and some run 880nm. Omnilux's Contour line at 633 and 830nm is the closest match we found among home devices, and FlexBeam's near-infrared peaks at 814nm. Whether the 20nm gap matters clinically has not been tested, so this is a tiebreaker between similar devices rather than a reason to reject a good one.

Can red light therapy make a scar worse?

Overdosing is the failure mode that actually appears in the literature. Photobiomodulation has a biphasic dose response, and in Kurtti's dose-ranging trial the highest cumulative fluence arm at 480 J/cm² improved induration by 50.0% against 71.4% on the untreated side — worse than doing nothing — while producing the only blistering in the trial set. Longer is not better, and the trials that worked used 4 to 13 J/cm² per session, which most consumer panels deliver in one to three minutes. The other way to make things worse is neglecting the treatments that do have first-line evidence: silicone sheeting, tension reduction and sun protection on a new scar all matter more than any panel.

---

Our Verdict

If we were treating a scar today, we would buy silicone sheeting first, keep the scar out of the sun, and add light as the cheap extra rather than the plan. For the light part we would take the Hooga HG300 at around $199: non-contact, 660 and 850nm, and enough published irradiance that a trial-sized 4.5 J/cm² session lands at roughly a minute at six inches. It is the smallest sensible spend on the least certain part of the protocol, which is the correct way round.

For a scar on the face, the Omnilux Contour Face at around $395 is the better-reasoned buy despite the price. Its 633 and 830nm sit exactly on the wavelengths the trials used, the ten-minute program removes the temptation to overdose, and a mask holds a constant geometry that a hand-held panel never does. For wrapping a limb or a C-section scar once it is fully mature, the Hooga belt at around $249 has a one-minute timer floor, which is the most trial-faithful dosing control on any device here.

What we would not do is expect much. The strongest home-use evidence is one 43-patient trial whose subjective endpoints moved and whose objective imaging did not, and the honest expectation is a scar that ends up slightly better than it would have been — not one that goes away.