Irradiance is the headline number in this category. It sits on the box, it drives the comparison charts, and it is the main thing brands compete on.
It is also the least reliable figure on any red light spec sheet, and the reasons are not subtle. There is no standard method, no independent verification, and no penalty for picking the measurement conditions that produce the biggest number.
What the number is supposed to mean
Irradiance is power per unit area, written in milliwatts per square centimeter. It tells you how intensely the device is lighting a given patch of skin.
Multiply it by time and you get dose, in joules per square centimeter, which is the quantity the research actually reports. Dosage walks through the conversion.
For that arithmetic to work, the irradiance figure has to describe your session. Most published figures do not.
This is a known problem in the research, not just in marketing
The field has been complaining about this internally for years.
Hadis and colleagues published a review in Lasers in Medical Science in 2016 titled, with some feeling, "The dark art of light measurement". They identified three recurring failures across the photobiomodulation literature: irradiation parameters that are incomplete, inaccurate or unverified; miscalculation of dose; and misuse of the basic terminology for light properties.
Jenkins and Carroll made a similar argument in Photomedicine and Laser Surgery in 2011, noting that researchers frequently make critical errors and omissions when reporting parameters, and that journals publish studies with missing data, arithmetic errors and no verification of beam parameters at all. They listed eight things that should always be reported: wavelength, power, irradiation time, beam area at the skin surface, pulse parameters, anatomical location, number of treatments and interval between treatments.
If peer-reviewed papers get this wrong routinely, a marketing department has no chance and no incentive.
Reason one: the meter
Most brands measure with a solar power meter. These are cheap, handheld, and designed for solar panel installers.
The problem is spectral response. A solar meter has a broadband detector that responds across a wide range of the spectrum and sums everything it receives into one figure. Point it at a red light panel in a normally lit room and it counts your therapeutic 660nm, your 850nm, whatever ambient light is around, reflected light off nearby surfaces, and infrared heat.
A calibrated spectroradiometer or a wavelength-specific optical power meter separates those out and reports what is actually arriving at the wavelengths you care about. Reviewers who test with proper equipment consistently record lower numbers than the ones on the box, often by a factor approaching two.
Reason two: distance goes unstated, or is zero
Irradiance falls off fast with distance. NovaaLab publishes a curve for its Deep Healing Pad that makes the shape clear: roughly 150 mW/cm2 at contact, 120 at one inch, 45 at three inches, 25 at six inches.
Moving from contact to six inches costs about five sixths of the intensity. So a figure measured at the panel surface, or at a distance the brand does not disclose, is not a description of anything you will experience.
Joovv rates the Solo 3.0 at greater than 100 mW/cm2 at the surface. Independent measurement at 15cm found roughly 60 to 90 mW/cm2 with a peak around 74. Both numbers can be honest. Only one describes a session.
Reason three: peak instead of average
LED panels are not uniformly bright. Output is highest directly in front of the densest cluster of diodes and falls off toward the edges and in the gaps between them.
A brand that reports the single hottest reading on the panel face is reporting a peak. A brand that reports the average across the treatment area is reporting something useful. The gap between the two is routinely 15 to 30 percent on the same device.
Independent testing of the Mito Red MitoPRO 1500+ found about 87 mW/cm2 peak and 76.5 mW/cm2 average. That is a well-behaved device with a small spread, and the difference is still material if you are computing session times.
Reason four: adding wavelengths together
A panel running 660nm and 850nm at once can be measured as two separate outputs or as one combined figure. Combining them roughly doubles the headline.
That would be defensible if red and near-infrared were interchangeable. They are not. They reach different depths and the clinical protocols specify them separately. See wavelengths.
On a seven-wavelength panel this gets worse, because the total is spread across bands with very different amounts of supporting evidence, and the combined number implies more useful light than is actually landing at any one wavelength.
Reason five: nobody has to verify anything
There is no required test standard for consumer irradiance claims, no third-party audit, and no regulator checking the box. FDA clearance does not cover this either. Clearance means a device was found substantially equivalent to an existing device for a stated use, which is not the same as FDA approval and says nothing about output claims. Most panels are only FDA registered, which is an administrative listing.
The result is a market where the incentive runs one way. PlatinumLED advertises 300+ mW/cm2 at six inches for the BIOMAX 900, which sits far above what independent testers record for panels in that class. The hardware is good. The number is a best-case laboratory figure being used as a comparison metric.
Why this matters less than you would think
Here is the part that reframes the whole argument: irradiance is not the variable that determines whether these devices work.
The sham-controlled mask trial by Park, Park and Jung in Medicine (Baltimore) in 2025 used a device capped at 10 mW/cm2 and reported crow's feet improvement in 86.2 percent of the active group against 16.7 percent of sham. The hair helmet in Yoon and colleagues' sham-controlled trial in Medicine (Baltimore) in 2020 measured 2.56 mW/cm2 and produced a 41.90 hairs per square centimeter density gain.
Those are tiny numbers next to any panel on the market. They worked because the total dose and the treatment schedule were right.
Photobiomodulation follows a biphasic dose response, described by Huang, Chen, Carroll and Hamblin in Dose Response in 2009. Below a threshold you get nothing, above the useful window the effect declines. Higher irradiance buys shorter sessions, which is a genuine convenience benefit. It does not buy better results, and past a point it works against you.
What a credible number looks like
A trustworthy irradiance claim states four things: the wavelength or wavelengths it refers to, the distance it was measured at, whether it is a peak or an average across the treatment area, and what instrument was used.
Almost nobody publishes all four. When a brand publishes none of them, treat the figure as marketing and go looking for an independent measurement instead.
Our device pages label every irradiance figure as either a brand claim or an independent measurement, because the gap routinely runs to a factor of two.
FAQ
Should I buy a solar power meter and test my own device? It will tell you whether your panel is degrading over time, which is useful. It will not give you an accurate absolute figure, because the detector responds to far more than your therapeutic wavelengths.
If the claim is inflated, are my sessions too short? Possibly. If you calculated a session length from a brand figure and the real output is half of it, you are getting half the intended dose. Assume the real number at a normal working distance is meaningfully lower and lean slightly longer, without going far beyond the ranges in published protocols.
Is a lower-irradiance device worse? Not necessarily. It needs longer sessions for the same dose, and it may be closer to what the trials actually used. Contact devices with modest output frequently deliver more light to the target than a high-output panel used at distance. Panel, mask, wrap or cap covers this.
Why do brands not just publish honest numbers? Because the market rewards the biggest figure and there is no cost to overstating it. Until independent testing becomes routine, the incentive will not change.
Sources
- Hadis MA, Zainal SA, Holder MJ, Carroll JD, Cooper PR, Milward MR, Palin WM. The dark art of light measurement: accurate radiometry for low-level light therapy. Lasers in Medical Science. 2016;31(4):789-809. https://link.springer.com/article/10.1007/s10103-016-1914-y
- Jenkins PA, Carroll JD. How to report low-level laser therapy (LLLT)/photomedicine dose and beam parameters in clinical and laboratory studies. Photomedicine and Laser Surgery. 2011;29(12):785-787. https://pubmed.ncbi.nlm.nih.gov/22107486/
- Huang YY, Chen AC, Carroll JD, Hamblin MR. Biphasic Dose Response in Low Level Light Therapy. Dose Response. 2009;7(4):358-383. https://pubmed.ncbi.nlm.nih.gov/20011653/
- Park SH, Park SO, Jung JA. Clinical study to evaluate the efficacy and safety of home-used LED and IRED mask for crow's feet: a multi-center, randomized, double-blind, sham-controlled study. Medicine (Baltimore). 2025;104(7):e41596. https://pmc.ncbi.nlm.nih.gov/articles/PMC11835066/
- Yoon JS, Ku WY, Lee JH, Ahn HC. Low-level light therapy using a helmet-type device for the treatment of androgenetic alopecia: a 16-week, multicenter, randomized, double-blind, sham device-controlled trial. Medicine (Baltimore). 2020;99(29):e21181. https://pmc.ncbi.nlm.nih.gov/articles/PMC7373546/