Treatments and gadgets. 11 minute read.
Red light therapy for joint pain: what the trials show and what the adverts claim
By the PocketPhysio editorial team. Published 2026-10-01.
The name on the box, and the name in the journals
Red light therapy is a retail term. The research you want is filed under photobiomodulation, and before about 2015 under low level laser therapy, which is why searching the consumer phrase turns up mostly people with something to sell.
The change of name was not cosmetic. Lasers turned out not to be necessary, since non-coherent light-emitting diodes did the same job in studies; low level was a subjective label that nobody could define; and the old name could not describe a treatment in which damping biological processes down and stimulating them can both be useful (Hamblin, 2016). So one phrase now stretches from a clinic laser unit to a panel, a wrap, a mask or a belt bought online, and it tells you nothing about what reaches your knee.
That matters for a practical reason, not a pedantic one. Every useful figure in this field is attached to a wavelength and a number of joules delivered to a point on the skin. Strip those off and you have a color.
What the knee osteoarthritis trials found
The knee is where almost all of the joint evidence sits. A systematic review of 22 randomized placebo-controlled trials, with data from 1,063 participants in total, found pain 14.23 mm lower than placebo on a 100 mm visual analogue scale at the end of the treatment course (95% confidence interval 7.31 to 21.14), pooled across the 816 people with an end-of-treatment pain score, and 15.92 mm lower in the 581 followed up 1 to 12 weeks later. Disability also improved, with a standardized mean difference of 0.59 (0.33 to 0.86) in 617 people. No adverse events were reported by any of the trial authors (Stausholm and colleagues, 2019).
Is 14 mm on a 100 mm scale a difference you would feel? The same paper puts three published estimates of the smallest improvement patients notice at 19.9, 17 and 9 units on that scale, depending on the study and on how much pain people started with. The pooled effect sits inside that spread rather than clearly above it.
Even that comparison is rough. Those thresholds describe an improvement within one person, and they shift with the pain that person started with, as the review itself points out, while 14.23 mm is the average gap between two groups.
In the subgroup of trials using the doses the field recommends the gap was larger, at 18.71 mm (9.42 to 27.99) in 480 people, and larger again at follow-up.
The second review is the more directly useful one for anyone already exercising. Fourteen trials in 820 people compared laser plus exercise therapy against placebo laser plus the same exercise therapy, which isolates what the light adds. Pain was lower in the laser group immediately after the course (standardized mean difference 0.58) and at follow-up between 4 and 32 weeks (1.35), the second of those at a p value of 0.05, which is as borderline as a result gets (Malik and colleagues, 2023).
The dose is the whole argument
Most pages on this subject leave the dose out, and it explains nearly all of the disagreement. In those 22 trials, the energy delivered to each treatment point ranged from 0.054 J to 8 J. Mean output power ranged from 0.3 mW to 400 mW. Course length ranged from 5 to 16 sessions, and 9 of the 22 trials used doses outside what the field itself recommends (Stausholm and colleagues, 2019).
That is a spread of more than a hundredfold in the quantity most likely to matter. Unsurprisingly, the trials did not agree with each other: statistical heterogeneity was 93% for the pooled pain analysis and 95% within the recommended-dose subgroup. A pooled figure built on that much disagreement is a summary of a messy literature, not a reliable prediction for one person.
The two reviews each end with a dose. The 2019 one concludes that pain and disability fell at 4 to 8 J per point with a 785 to 860 nm laser, and at 1 to 3 J per point with a 904 nm laser, so the figure that counts as enough depends on the wavelength (Stausholm and colleagues, 2019). The 2023 one reached a similar place from the other direction, concluding that pain fell at 4 to 8 J per point, at wavelengths of 640 to 905 nm, over 10 to 16 sessions at two a week (Malik and colleagues, 2023). Those are the numbers to compare a treatment against. If a clinic or a device cannot tell you its wavelength and its joules per point, nobody can say whether it resembles the trials or not.
What it did not change
The 2023 review is blunt about this, and this is the finding to carry away. Alongside its pain result it found no significant difference between laser plus exercise and placebo laser plus exercise for knee range of motion, for muscle strength, or for function, either immediately or at follow-up (Malik and colleagues, 2023).
So the honest summary is a treatment that may reduce pain for a while and does not make a knee stronger, bend further or work better. The strengthening still has to be done. That is what moves those three.
Nothing in either review measured cartilage, joint space, imaging or how the arthritis progressed. The outcomes were pain, disability, range of motion, strength and function. There is therefore no support anywhere in this evidence for the idea that light repairs or rebuilds a joint, and anyone who tells you otherwise is not drawing on these reviews.
Other joints and tendons
Outside the knee the picture thins quickly. For Achilles tendinopathy there is a systematic review of 4 trials in 119 participants, all of them adding laser to eccentric exercise and comparing it against sham laser with the same exercise. It rated the certainty of the evidence very low to low. One trial favored laser at two months, but at 3 and 13 months the confidence intervals did not include a difference worth having, and the function score at one month favored the placebo group. The authors concluded the evidence does not support routine use (Martimbianco and colleagues, 2020).
Neck pain is the other application with a substantial trial base. It is where the one systematic review of this treatment in The Lancet sits: 16 trials in 820 patients, with pain reduced immediately after treatment in acute neck pain and still lower up to 22 weeks after the course in chronic neck pain. The pooled reduction was 19.86 mm on a 100 mm scale (95% confidence interval 10.04 to 29.68) in the 11 of those trials that reported a change on that scale (Chow and colleagues, 2009). That review was published with a linked commentary and drew correspondence in the same journal the following year, which is normal for a contested result and worth knowing if someone quotes it as settled.
For the hand, the hip, the shoulder or the back, this page has nothing to offer, because no review in those areas was found while it was written. That is an absence of evidence rather than evidence of absence, and it is also a reason not to spend money on the basis of a knee result.
Where the guidelines sit, and why the laser researchers disagree
Guideline bodies have generally not been persuaded, and the pro-laser side says so itself. The 2019 review opens with the line that laser therapy is not recommended in the major osteoarthritis guidelines: it was mentioned but not recommended in the 2018 European League Against Rheumatism guidelines, and the 2018 OARSI guidelines said it should not be treated as a core intervention for knee osteoarthritis (Stausholm and colleagues, 2019). The Australian guideline for knee and hip osteoarthritis makes a conditional recommendation against laser therapy, while strongly recommending land-based exercise for knee osteoarthritis and weight management for people who are overweight (Royal Australian College of General Practitioners).
What happened next is unusual: the recommendation was publicly contested in print. A narrative review in Lasers in Medical Science argued that the guideline's search had picked up 8 trials when around 25 existed, that the most recent trial it cited was from 2012 in a guideline published in 2018, that massage was recommended on the same low level of evidence that laser therapy was recommended against, and that nobody with expert knowledge of laser dosing had been consulted (Chow and colleagues, 2021). Who wrote what is worth knowing here. The first author of that narrative review is the first author of the 2009 neck pain review, and two of the authors of the 2019 meta-analysis it relies on are co-authors of that same 2009 review, so these are not independent voices checking each other. All of them declared no conflict of interest.
Both positions can be read charitably. If dose is the deciding variable, then pooling trials that used doses a hundredfold apart will dilute a real effect, and a guideline that does the pooling will find little. If dose is not the deciding variable, the subgroup analyses are the kind of result that often fails to replicate. That argument is not resolved, and a page that pretended it was would be the fourth problem in your search results, not the answer.
The claims you cannot check
Search the consumer phrase and the first page is mostly device sellers. Among the claims they make, one pattern deserves naming: a systematic review attributed to The Lancet, said to show a large benefit for pain, with no authors, no year, no title given, and a specific percentage attached to it.
That claim does not survive a search. The only systematic review of this treatment published in The Lancet that could be found while this page was written is the 2009 neck pain review described above, which is not about joint repair, is not about arthritis, and does not match the claim being made. The number attached to the claim is not repeated anywhere on this page, because a figure with no traceable paper behind it is not evidence, and repeating it even to argue with it is how it spreads.
Checking costs nothing. A real citation names its authors and its year and the journal it appeared in, so you can go and read it. A real dose claim gives a wavelength in nanometers and an energy in joules per point, not a number of minutes. And a claim about rebuilding cartilage has no support in the reviews on this page, whatever the picture next to it shows.
Is it safe?
As far as the trials go, uneventfully so. Across the 22 placebo-controlled knee trials pooled in 2019, no adverse events were reported by any of the trial authors (Stausholm and colleagues, 2019).
That is reassuring and it is not a safety study. Those trials were short, delivered in clinics, at stated doses, to knees, in people who had been screened to enter a trial. They say nothing about long daily use of a home device, or about skin that has something else going on, and the instructions for any device are the maker's to give and yours to follow.
One caution comes out of the numbers, not out of the trials. The clinic units in those 22 trials ran at mean output powers up to 400 mW. OSHA puts continuous lasers of 5 to 500 mW in hazard class 3B and describes that class as an immediate eye hazard when the beam is viewed directly, so a 400 mW clinic laser sits inside it (OSHA Technical Manual; OSHA laser hazards). Most consumer panels and wraps use light-emitting diodes rather than lasers, so that classification does not transfer to every device on sale, but the sensible rule covers both: point it at the joint, keep it away from your eyes, and use any eye shields supplied with it. None of the trials here measured eye exposure, so that is caution rather than a finding.
Two more cautions belong in the same bracket. If you are having treatment for cancer, ask your cancer team before using a light device over any area they are treating or watching. And if the skin you would be treating is broken, infected or has a mole or patch that has changed, get that looked at before you shine anything on it.
The point that matters clinically is about delay. A joint that is hot, red and swollen, or painful alongside feeling feverish or shivery, needs to be seen the same day rather than treated at home with anything, and pain that wakes you at night and keeps getting worse needs a doctor within a few days. Both are in the warning list below.
Where it fits beside the exercise
Treat it as an add-on and the evidence becomes easy to read. Eleven of the 22 trials in the 2019 review used laser alongside exercise therapy, and both of the reviews here that tested it as an addition still found a pain effect. The fairest version of its usefulness is that it may make the painful weeks easier while the exercise does the work that changes strength and function.
The work itself has stronger backing, and it is what this site is built around. For an arthritic knee, start with the knee osteoarthritis program; for the hip, hip osteoarthritis; for sore, stiff finger joints, hand osteoarthritis; and for inflammatory joint disease, rheumatoid arthritis, where your rheumatology team's plan comes first. A sore neck has its own staged program at neck pain, and a knee that hurts on stairs and after sitting is covered by knee pain on stairs.
If you want to try a course, the sensible way to do it is to keep the exercise going unchanged, agree in advance what you are judging and when, and remember that in the recommended-dose trials of the 2019 review the largest difference from placebo arrived 2 to 4 weeks after the sessions stopped.
When to see a physio (physical therapist)
Before you buy anything, if the joint has not been assessed. Most of the money spent on devices for joint pain is spent before anyone has worked out which structure is sore, what makes it worse and what it can tolerate, and that assessment changes the plan far more than any gadget does.
Book sooner if a joint has been painful for more than a few weeks, if it is swelling after activity, if you have started avoiding stairs or hills, or if the pain has begun waking you. Take a note of what you have already tried, including any light therapy, how long for, and what changed. The warning list below comes first, at the speed each line sets.
See a doctor promptly if
- Same day: a joint is hot, red and swollen, or you have joint pain with a high temperature or feel hot, cold or shivery. Get medical advice the same day, and go to an emergency department if you feel very unwell. This can be a sign of infection in the joint, which needs treatment quickly.
- Same day: a calf or thigh that is newly swollen, warm, tender, red or darker than usual, has swollen veins that are sore to touch, or has a throbbing or cramping pain that does not fit your injury or condition, or that feels different from normal muscle soreness after exercise. This can be a sign of a blood clot (deep vein thrombosis), especially after surgery, an injury, time in a cast or boot, a long trip or a spell of being much less mobile than usual. Get medical advice the same day. If you are also short of breath or have chest pain, call emergency services.
- Within a few days: the pain is there at night or at rest and keeps getting worse. See your doctor within a few days. If you have had cancer, now or in the past, or are losing weight without trying, see your doctor sooner, within a day or two, and mention it. This applies even if the pain is not getting worse. If you are being treated for cancer now, contact your cancer team the same day.
Common questions
Does red light therapy work for joint pain?
For knee osteoarthritis pain in the short term, the better evidence says yes, modestly, and the same evidence says it changes nothing else. A systematic review of 22 placebo-controlled trials, 1,063 participants in all, found pain 14.23 mm lower than placebo on a 100 mm scale at the end of treatment in the 816 who had that measurement (95% confidence interval 7.31 to 21.14), rising to 18.71 mm in the 480 from the subgroup of trials that used the doses the field recommends (Stausholm and colleagues, 2019). A separate review of 14 trials in 820 people, all of them comparing laser plus exercise against placebo laser plus exercise, found less pain and no difference at all in range of motion, muscle strength or function (Malik and colleagues, 2023). Both sets of trials were short, the pooled trials disagreed with each other a great deal, and neither review looked at any joint other than the knee; the same 2019 review opens by saying that laser therapy is not recommended in the major osteoarthritis guidelines. Nothing in either review measured cartilage, joint space or how the arthritis progressed, so none of it says a joint is repaired.
How does red light therapy for pain work?
The mechanism is not settled, and the field's own history says so. The name changed from low level laser therapy to photobiomodulation partly because lasers turned out not to be necessary, partly because nobody could define what low meant, and partly because the old name could not cover the fact that damping biological processes down can be as useful as stimulating them (Hamblin, 2016). What the outcome reviews measure is whether pain scores move, not why. Anyone offering you a confident cellular explanation is going beyond the evidence that is actually being cited.
How long should you do red light therapy for pain?
The honest answer is that this question cannot be answered from the trials for a home device, because the trials describe their dose in units that device boxes rarely print. In the 22 trials pooled in 2019, energy per treatment point ranged from 0.054 J to 8 J, mean output power from 0.3 mW to 400 mW, and courses from 5 to 16 sessions (Stausholm and colleagues, 2019). The 2023 review concluded that pain was reduced at 4 to 8 J per point at a wavelength of 640 to 905 nm, applied for 10 to 16 sessions at two sessions a week (Malik and colleagues, 2023). Those are clinic protocols delivered by someone who knows the output of the device, and minutes on a timer is not the same quantity.
How often should you do red light therapy for pain?
Twice a week for 10 to 16 sessions is what the 2023 review named for knee osteoarthritis, and the trials in the 2019 review ran courses of 5 to 16 sessions (Malik and colleagues, 2023; Stausholm and colleagues, 2019). In the 2019 review the effect on pain did not peak at the last session: in the trials using recommended doses, the largest difference from placebo came 2 to 4 weeks after the course had finished, at 31.87 mm on a 100 mm scale in the 322 people with data at that point (95% confidence interval 18.18 to 45.56). Both of those session counts come from clinic trials in which someone knew the output of the device, so they do not translate into minutes on a home timer. If you are considering a course, the timing is the thing to ask about, because judging it on the day of the final session may be judging it too early.
Does red light therapy work for back pain?
No systematic review of this treatment for low back pain was found while this page was written, so the question is left open here rather than answered. What exists in the sources on this page is knee osteoarthritis, where the pain effect is short-term and modest, neck pain, where a 2009 review in The Lancet reported pain reduced immediately after treatment in acute neck pain and up to 22 weeks later in chronic neck pain, and Achilles tendinopathy, where a review of 4 trials in 119 people rated the certainty of the evidence very low to low (Chow and colleagues, 2009; Martimbianco and colleagues, 2020). For a sore back, the low back pain program is the part with guideline support.
Is red light therapy at home the same as in a clinic?
Not in any way you can check, which is the problem. The trials specify a wavelength in nanometers, an energy in joules delivered to a named number of points on the joint, and a session count; a home panel or wrap usually gives you a wattage, a distance and a timer. Without the wavelength and the joules per point you cannot tell whether a device sits inside the range the reviews found effects at, or a hundredfold outside it. That is not a claim that home devices do nothing. It is a statement that nobody, including the seller, can map one onto the other from the information on the box.
References
- Stausholm MB, Naterstad IF, Joensen J, Lopes-Martins RAB, Saebo H, Lund H, Fersum KV, Bjordal JM. Efficacy of low-level laser therapy on pain and disability in knee osteoarthritis: systematic review and meta-analysis of randomised placebo-controlled trials. BMJ Open. 2019;9(10):e031142. https://doi.org/10.1136/bmjopen-2019-031142
- Malik S, Sharma S, Dutta N, Khurana D, Sharma RK, Sharma S. Effect of low-level laser therapy plus exercise therapy on pain, range of motion, muscle strength, and function in knee osteoarthritis: a systematic review and meta-analysis. Somatosensory and Motor Research. 2023;40(1):8-24. https://doi.org/10.1080/08990220.2022.2157387
- Hamblin MR. Photobiomodulation or low-level laser therapy. Journal of Biophotonics. 2016;9(11-12):1122-1124. https://doi.org/10.1002/jbio.201670113
- Chow RT, Johnson MI, Lopes-Martins RA, Bjordal JM. Efficacy of low-level laser therapy in the management of neck pain: a systematic review and meta-analysis of randomised placebo or active-treatment controlled trials. The Lancet. 2009;374(9705):1897-1908. https://doi.org/10.1016/S0140-6736(09)61522-1
- Guzman J. Neck pain and low-level laser: does it work and how? The Lancet. 2009;374:1875-1876. Linked comment. https://doi.org/10.1016/S0140-6736(09)61837-7
- Shiri R, Viikari-Juntura E. Low-level laser therapy for neck pain. The Lancet. 2010;375:721-722. Correspondence, with author reply. https://doi.org/10.1016/S0140-6736(10)60297-8
- Verhagen AP, Schellingerhout JM. Low-level laser therapy for neck pain. The Lancet. 2010;375:721. Correspondence, with author reply. https://doi.org/10.1016/S0140-6736(10)60296-6
- Martimbianco ALC, Ferreira RES, Latorraca COC, Bussadori SK, Pacheco RL, Riera R. Photobiomodulation with low-level laser therapy for treating Achilles tendinopathy: a systematic review and meta-analysis. Clinical Rehabilitation. 2020;34(6):713-722. https://doi.org/10.1177/0269215520912820
- Royal Australian College of General Practitioners. Guideline for the management of knee and hip osteoarthritis. Summary: plain language. Accessed 1 October 2026. https://www.racgp.org.au/clinical-resources/clinical-guidelines/key-racgp-guidelines/view-all-racgp-guidelines/knee-and-hip-osteoarthritis/summary-plain-language
- Chow R, Liebert A, Tilley S, Bennett G, Gabel CP, Laakso L. Guidelines versus evidence: what we can learn from the Australian guideline for low-level laser therapy in knee osteoarthritis? A narrative review. Lasers in Medical Science. 2021;36(2):249-258. Published online 8 August 2020. https://doi.org/10.1007/s10103-020-03112-0
- Occupational Safety and Health Administration. Laser hazards: hazard recognition. Accessed 1 October 2026. https://www.osha.gov/laser-hazards/hazards
- Occupational Safety and Health Administration. OSHA Technical Manual, Section III Chapter 6: Laser Hazards. Accessed 1 October 2026. https://www.osha.gov/otm/section-3-health-hazards/chapter-6
- NHS. Septic arthritis. Page last reviewed 3 September 2026. https://www.nhs.uk/conditions/septic-arthritis/
Written and checked by the PocketPhysio editorial team. Last updated 2026-10-01.
This page is general education, not a diagnosis or a personal treatment plan. See a physiotherapist or doctor for advice about your own situation.