Red light therapy for hair loss: what the sham-controlled trials show

Hair is the one place in red light where the evidence is mostly randomised, sham-controlled and measured in a hard unit: hairs per square centimetre, counted by a blinded assessor on a photograph of a tattooed spot of scalp. What it supports is narrower than what the caps’ marketing claims.

Does red light therapy work for hair loss?

For pattern hair loss, the sham-controlled trials find a real but modest effect. The ones worth knowing, with the design in each case:

• Jimenez et al., 2014. Four multicentre, randomised, double-blind, sham-controlled trials. 269 people were randomised, 128 men and 141 women with pattern hair loss, using a HairMax laser comb (7, 9 or 12 beams) on the whole scalp three times a week for 26 weeks. Terminal hair density rose by 18.4 to 25.7 hairs per cm² in the laser groups, against 1.6 to 9.4 in the sham groups. No serious adverse events.
• Lanzafame et al., 2013 and 2014. Two double-blind RCTs of a helmet holding 21 laser diodes and 30 LEDs at 655nm, used at home for 25 minutes every other day for 16 weeks. The sham helmet held incandescent red bulbs. In 44 men (Norwood IIa–V), the hair count in a 2.85 cm² tattooed patch rose by 67 in the active group and 32 in the sham group. In 47 women (Ludwig I–II), it rose by 100 against 24.
• Kim et al., 2013. A 24-week double-blind RCT of 40 people using a home helmet at 630, 650 and 660nm for 18 minutes daily. Hair density and thickness improved more than with sham. The investigators’ global assessment favoured the device; the subjects’ own assessment did not.

The meta-analyses agree on direction. A 2019 analysis of 11 double-blind RCTs and a 2021 analysis of 15 studies (795 people) both found standardised effects of about one standard deviation in favour of light over sham. A 2025 review of 38 studies (3,098 patients, almost all with androgenetic alopecia) found the same, with heterogeneity above 80%, meaning the trials disagree a lot on how big the effect is. A 2018 network meta-analysis of 22 studies ranked light therapy first of six non-surgical treatments, minoxidil and finasteride included, on mean change in hair count. It also rated the evidence for light against sham very low quality.

None of them shows what happens over years, or after you stop.

Red light therapy can’t fix hair loss with another cause

Light therapy for hair has been tested almost entirely on one condition. The 2025 review above found 2,930 of its 3,098 patients had androgenetic alopecia. It had 49 with scarring alopecia, and too little data on those to analyse.

That matters because “hair loss” is several different problems:

• Scarring alopecias (lichen planopilaris, frontal fibrosing alopecia, central centrifugal cicatricial alopecia) replace the follicle with scar tissue. There is nothing left for light to act on, and delay costs hair that does not come back.
• Telogen effluvium is diffuse shedding after illness, childbirth, surgery, crash dieting, iron deficiency, thyroid disease or a new medication. It is treated by finding the cause.
• Alopecia areata is autoimmune, usually in round patches, and has its own treatments.

Is a red light hair cap FDA approved? What the clearance covers

No. Like light masks, home hair-growth devices reach the market through 510(k) clearance. The FDA finds the device substantially equivalent to one already sold for the same use. That is not approval. The mask guide explains the difference at more length.

The detail that matters for hair is the cleared indication. It is narrow, and it is in the public record. Take the cap picked below. The FDA’s 510(k) database lists the iRestore Professional 282 as K183417, cleared on 17 May 2019 for over-the-counter use. It is cleared to promote hair growth in women with androgenetic alopecia at Ludwig-Savin I–II, and men at Norwood-Hamilton IIa–V, with skin phototypes I to IV in both. Three things follow:

1. Advanced loss is outside the claim. Norwood VI–VII and Ludwig III are not in the indication. The trials did not enrol them.
2. Skin types V and VI are outside it too. That matches who was enrolled: the Lanzafame trials took Fitzpatrick I–IV only.
3. The clearance relied on an earlier device. The 510(k) summary says no clinical performance data was produced for that submission. Equivalence was argued from the original iRestore (K151662), which the same document lists as 21 laser diodes and 30 LEDs at 650nm. That matches the counts in the Lanzafame helmet. One of those trials’ authors is also the consultant named on the K183417 clearance letter. That is common in this field, and it is worth knowing when you weigh the trials.

The FDA lists dozens of cleared hair combs, caps and helmets under the same product code. Clearance does not tell you one beats another.

Laser cap vs LED cap for hair growth

Hair devices use laser diodes, LEDs, or a mix, and the box usually totals them into one number. The evidence suggests you should split them:

• A 2021 meta-analysis (Gupta and Carviel, 15 studies) found laser-only devices significantly more effective than laser-and-LED combinations. The combinations still beat sham. Comb versus helmet made no significant difference.
• A 2021 meta-analysis limited to FDA-cleared home devices (Lueangarun et al., seven RCTs) found the same pattern: a standardised effect of 1.52 for laser-only devices and 0.85 for combinations.

Both findings compare different trials of different devices, not a laser cap against an LED cap in one study. A 2014 review from the group behind much of this research says the best wavelength, coherence and dose have not been worked out. So what we can say is this: the best-supported devices are laser-heavy, and a diode count that does not separate lasers from LEDs hides the one split the evidence points to.

Best red light therapy for hair growth: what to demand from any cap

The cheaper caps we found on Amazon had few reviews, unfamiliar brands and unverifiable specs, so none is on this page. Ask any cap for these, in order:

1. A K-number, and an indication that matches your hair loss. Look it up in the FDA 510(k) database. If your pattern is outside the cleared Norwood or Ludwig range, the clearance does not cover you.
2. Lasers and LEDs counted separately. “282 lasers and LEDs” is one number covering two kinds of diode.
3. Wavelength around 650–660nm. The trials above used 630–660nm, and 655nm most often.
4. A protocol you will keep. The cleared and trialled protocols are 25 minutes every other day, or three times a week, for 16 to 26 weeks. A hands-free cap you actually wear beats a comb you put down after a fortnight.
5. Coverage where you are losing hair. Crown and hairline are different targets, and a comb treats only where you move it.
6. Diode count, last.

The device with the largest published sham-controlled trial is the HairMax laser comb from the Jimenez study. If you choose a comb, judge it on whether you will use it three times a week for six months.

Red light therapy hair growth: how long until results

Plan on four months before judging anything and six months before deciding. The trials measured at 16 weeks (Lanzafame, Yoon), 24 weeks (Kim, Fan) and 26 weeks (Jimenez), and the iRestore’s 510(k) states a 16-week course. Hair grows in cycles that run for months.

Three practical notes:

• Take baseline photos. Use the same light, parting, distance and hair length, at the crown and hairline. Without them, month four is guesswork.
• More often is not better. In the 2019 meta-analysis, trials with lower treatment frequency showed a larger effect than higher-frequency ones. That fits the biphasic dose response in the dosing guide. Follow the device’s schedule.
• Combining with minoxidil is unsettled. Two 2025 meta-analyses disagree. One, of seven RCTs, found 6.6 more hairs per cm² with laser plus minoxidil than minoxidil alone. The other, of four RCTs, found no difference.

Why the panel dose arithmetic does not transfer

Elsewhere on this site dose is simple: irradiance × seconds ÷ 1000 = J/cm². On a scalp it gets harder. The cap sits on or near the skin, but hair is in the way, and hair absorbs light before it reaches skin. The trials also report dose in ways that do not line up. The Lanzafame trials reported 67 J/cm² per 25-minute session. Taken at face value, that is about 45 mW/cm² (67,000 ÷ 1,500 seconds). A 2020 helmet trial (Yoon et al., 60 people, 16 weeks) reported a mean output of about 2.4 mW/cm² at 655nm, and still found an increase of 41.9 hairs per cm² against 0.7 with sham. Two positive trials are roughly twenty-fold apart on irradiance.

So we will not give you a J/cm² target for hair. The cleared protocol for your device is the dose control that has been tested.

Safety: eyes, scalp and medication

The trials above reported no serious adverse events. The rules are short:

• Do not look into the diodes. The iRestore’s 510(k) classifies its diodes as Class 3R lasers. Keep the cap on your head while it is running, and do not turn it over to see whether it is on.
• Not on broken, burnt or inflamed scalp. Get a scalp that is scaling, weeping or painful diagnosed first. See the warning above.
• Check photosensitising medication with a pharmacist. Photodynamic therapy drugs such as aminolevulinic acid and verteporfin are designed to be activated by red light. Several common antibiotics, retinoids and St John’s wort make skin more light-sensitive. We found no trial data in pregnancy.

What we have not measured

We have not put a hair cap on a bench, and we have not run a hair-growth trial. The trial results above come from the published papers, as stated in their abstracts. The diode counts, wavelengths and protocols come from the FDA 510(k) summary for K183417 or from the Amazon listing, and each sentence says which. No cap we looked at publishes irradiance at the scalp.

The test that would change this page is a spectroradiometer reading inside the cap. We would fit it to a head form and map irradiance at the scalp plane on a grid across the crown and hairline. We would read the laser diodes and the LEDs separately, with a calibrated laser power meter as a cross-check, and state the instrument’s calibration date. That gives mW/cm², how even the coverage is, and a real dose per session to compare across caps. It would be one unit of each model (n=1): our units, not every unit sold. What we will not do is photograph our own scalp for 16 weeks and call it evidence. One person’s hair counts cannot be told apart from natural variation, which is why the trials used sham devices. How we test says the same about every figure on the site.

Looking at red light for the face instead? The mask guide covers the same clearance question for skin.