If you have spent an hour reading about red light therapy, you have probably noticed something strange. Half the pages describe an unremarkable physical therapy adjunct with modest, inconsistent results. The other half describe something close to a universal remedy, good for everything from arthritis to brain fog. Both sets cite studies — sometimes the same studies.

This is the version a careful physician would give you: what photobiomodulation is, what is known, what is not, and how to tell the difference when reading a clinic’s website — including ours.

What it is

Photobiomodulation (PBM) is the use of red and near-infrared light, usually from lasers or LEDs, delivered to tissue at intensities too low to heat it. The same thing is called low-level laser therapy (LLLT), cold laser, or red light therapy. The names are largely historical; the idea is the same.

Nothing is cut, burned, or ablated. A surgical laser deposits enough energy to destroy tissue; a PBM device deposits far less, aiming at a biological response rather than a thermal one. That is also why the field is hard to study: when a treatment produces no sensation, there are many ways to get the dose wrong and no way for the patient to notice.

The mechanism, stated honestly

The leading mechanistic hypothesis is that red and near-infrared photons are absorbed by cytochrome c oxidase, an enzyme complex at the end of the mitochondrial electron transport chain. Absorption is proposed to alter the enzyme’s activity, with downstream changes in signaling molecules — reactive oxygen species, nitric oxide, calcium — that influence gene transcription related to inflammation and tissue repair.

That is a real line of laboratory research. It is also where marketing copy quietly changes gear, moving from "this happens in a cell culture" to "therefore this happens to your knee." Those are different claims requiring different evidence, and medicine’s history is largely a history of mechanistically elegant treatments that did nothing measurable in trials.

The mechanism is also dose-shaped: the laboratory literature describes a biphasic response, in which too little energy does nothing and too much can suppress the very response you were aiming for.

Wavelength, and how deep light actually goes

Light does not penetrate tissue uniformly. Blue and green wavelengths are absorbed heavily by hemoglobin and melanin within a fraction of a millimeter. Red light in roughly the 600–700 nm range reaches somewhat further. Near-infrared light, roughly 800–900 nm, passes through skin and superficial soft tissue better than either, which is why devices aimed at deeper structures such as tendons and joints tend to sit in that band. "Better" is relative, though: near-infrared still attenuates steeply, and what reaches a structure a few centimeters down is a small fraction of what left the emitter — an amount that shifts with skin pigmentation, overlying fat, and applicator contact.

The clearest illustration comes from lateral elbow tendinopathy. Pooling 13 trials and 730 patients, Bjordal and colleagues (2008, BMC Musculoskeletal Disorders) found overall pain relief of 10.2 mm on a 100 mm scale (95% CI 3.0–17.5, p=0.005) — unevenly distributed. Trials using 904 nm directly over the tendon showed 17.2 mm (95% CI 8.5–25.9, p=0.0001). Trials using 820 nm, 830 nm, or 1064 nm showed no significant effect, and trials irradiating acupuncture points instead of the tendon gave negative results.

Same modality, same condition, opposite conclusions — decided by wavelength and by where the light was pointed.

Dose is not a detail — it is the finding

"Does PBM work?" is nearly unanswerable, in the same way "does medication work?" is unanswerable without naming the drug and the dose. The best demonstration is knee osteoarthritis. Stausholm and colleagues (2019, BMJ Open) pooled 22 placebo-controlled trials, 1,063 patients. Averaged across everything, pain fell 14.23 mm on a 100 mm visual analog scale (95% CI 7.31–21.14) at end of therapy — an unremarkable figure.

The subgroup analysis is the interesting part. Trials using doses consistent with the World Association for PhotobiomodulaTion Therapy dosage recommendations showed 18.71 mm (95% CI 9.42–27.99), rising to a peak of 31.87 mm (95% CI 18.18–45.56) two to four weeks after the course had finished. Trials using non-recommended doses showed 6.34 mm, and that result was not statistically significant at follow-up. The difference between the two subgroups was itself significant, p = 0.02.

Two caveats belong alongside that. Heterogeneity was very high, I² = 93–95%: the trials disagreed with one another far more than chance explains. And subgroup analyses are hypothesis-generating, weaker than a trial built from the outset to compare two doses head to head. Oliveira and colleagues (2024, Physical Therapy), pooling 10 studies and 542 participants, found pain at rest improved (SMD −0.7, 95% CI −1.1 to −0.2) but rated that evidence very low certainty, concluding it "does not allow to recommend its isolated use but may be used to complement other widely recommended therapies."

The pattern repeats in the shoulder: Haslerud and colleagues (2015, Physiotherapy Research International) pooled 17 RCTs and reported that "trials performed with inadequate laser doses were ineffective across all outcome measures." A clinic that cannot state its wavelength, output, and energy per point cannot tell you whether it is delivering the version of this treatment that worked in the trials.

The honest overall picture

In 2025, Son and colleagues (Systematic Reviews) published an umbrella review — a review of systematic reviews — covering 15 meta-analyses, 204 randomized controlled trials, more than 9,000 participants, and 35 outcomes across 15 conditions. It is the broadest synthesis of PBM available.

The certainty ratings: 0 outcomes high certainty (0%), 6 moderate (17.1%), 20 low (57.1%), 9 very low (25.7%). In the authors’ words: "No outcome was supported by high certainty of evidence."

Six of the 35 outcomes were rated moderate; the review names five of them:

pain in burning mouth syndrome
disability in knee osteoarthritis (eSMD 0.65, 95% CI 0.14–1.15)
fatigue in fibromyalgia
hair density in androgenetic alopecia
cognitive function

Diabetic foot ulcers came out at very low certainty. Achilles tendinopathy and carpal tunnel syndrome showed no significant effect. The authors called for "standardization of PBM protocols… before widespread clinical adoption."

That is the state of play. The best-supported outcomes in the entire field sit at moderate certainty, meaning further research is reasonably likely to change the estimate. Nothing here is settled, and anyone writing as though it were is either not reading the literature or counting on you not to.

Where the evidence points the other way

A field is best judged by what its own researchers publish when the result is unfavorable.

Non-specific low back pain. Tomazoni and colleagues (2020, Journal of Physiotherapy) reviewed 12 RCTs, n=1,046, most at low risk of bias: "Current evidence does not support the use of PBMT to decrease pain and disability in people with non-specific LBP." Several of the authors are prominent PBM researchers, which makes the result hard to wave away.
Diabetic peripheral neuropathy. Lavery and colleagues (2008, Diabetes Care), a double-blind sham-controlled trial across 120 limbs over 90 days: "Anodyne MIRE therapy was no more effective than sham therapy."
Neck pain, where two respected reviews disagree — Chow 2009, Lancet versus Kadhim-Saleh 2013, Rheumatology International. We take that disagreement apart in Neck Pain and Low-Level Laser.

The strongest indication in the field, meanwhile, is one most people have never heard of. The MASCC/ISOO clinical practice guidelines (Zadik 2019, Supportive Care in Cancer; Elad 2020, Cancer) recommend PBM at Level I evidence to prevent oral mucositis in adults having hematopoietic stem cell transplantation with high-dose chemotherapy, and in head and neck cancer patients having chemoradiotherapy. Note the scope: prevention, in defined populations. For mucositis already established, the guideline states that no guideline was possible.

What photobiomodulation is not

It is not a replacement for anything. In the trials where it performed best, it sat on top of exercise, physical therapy, standard medication, or oncology supportive care. Adjunctive is not a hedge word here; it describes how the evidence was generated.

It does not make an underlying condition go away. Osteoarthritis is still osteoarthritis after a course of laser; what the better trials report is lower pain scores over weeks to months in groups of people, with wide variation between individuals.

It is not a single technology, either. A handheld LED panel, a 904 nm pulsed diode applied point by point over a tendon, and an intranasal device differ as much as three drugs that share a chemical family. Evidence for one does not transfer.

How to read a clinic’s claims — including ours

Do they name the parameters? Wavelength in nanometers, output in milliwatts, energy per point in joules, number of points, session count. If a page describes benefits at length and never mentions a number, the omission is the message.
Do they state the certainty, not just the effect? "SMD 0.66" means little without "and that estimate was rated low certainty."
Do they cite anything negative? A modality with 204 randomized trials behind it necessarily has unfavorable ones. A page carrying only supportive citations has been curated, not summarized.
Do they stretch a narrow clearance into a broad claim? The HairMax LaserComb holds FDA 510(k) clearance K060305, and its indications for use read, verbatim: "The LaserComb is indicated to promote hair growth in males with androgenetic alopecia who have Norwood Hamilton Classifications of IIa to V and Fitzpatrick Skin Types I to IV." One device, one population, one outcome — and clearance is not approval.
Do they promise you, personally, a result? Studies report group averages. A page that tells you what you will experience has stopped describing evidence.
Do they pair an unrecognized diagnosis with an unapproved device? That combination is the most reliable warning sign in this corner of medicine.

Where this leaves you

Photobiomodulation is a real intervention with a real, if uneven, evidence base. For certain musculoskeletal presentations — delivered at doses consistent with published recommendations, alongside an active rehabilitation plan — the pooled data show meaningful average pain reduction over a period of weeks. For other conditions the data are weak, mixed, or plainly negative, and we would rather say so before you spend money than after.

The useful questions are narrow: what is the target tissue, how deep is it, what wavelength and dose reach it, what does the evidence show for that application, what else should be happening alongside it, and how will we decide in six weeks whether it is doing anything for you. A clinician who cannot answer those is not in a position to recommend the treatment.

This article is general education, not medical advice, and it is not a substitute for evaluation by a qualified clinician. It does not establish a physician–patient relationship. Individual results vary, and the studies described report group averages rather than what any one person should expect. Treatments discussed may not be FDA-cleared or approved for the conditions mentioned; where that is the case it is stated in the text. Talk to your own physician before starting, stopping or changing any treatment.

What Photobiomodulation Actually Is — And What the Evidence Really Says