How does red light therapy work?

Red light therapy is simple to describe: red and near-infrared light shone on the skin for a few minutes at a time. What it does once it gets there is a cell-level story, and most of it happens in the mitochondria. This post is the short version of our How RLT Works page, with the studies linked as we go.

Where the light goes

Light only does something if a molecule absorbs it. The one that matters here is cytochrome c oxidase, an enzyme in the mitochondria that sits at the end of the chain that makes ATP, the cell's energy currency. The NASA LED research group described this twenty years ago: red to near-infrared light (630-1000nm) is absorbed by cytochrome c oxidase, and LED light increased the enzyme in cultured nerve cells. A 2018 review of the mechanism puts it the same way: the primary site of absorption in mammalian cells is the mitochondria, and more specifically cytochrome c oxidase.

What happens in the mitochondria

The working hypothesis, in that same review, is that light knocks inhibitory nitric oxide off the enzyme, restoring electron transport and raising the mitochondrial membrane potential. The word hypothesis is the authors' own - the absorption is measured, the nitric oxide step is the best current explanation of what follows.

The downstream part has been measured directly. In mouse muscle cells, a 90-second dose from a 630 and 850nm LED cluster at 28 mW/cm² raised membrane potential and ATP, peaking three to six hours after the light - cells in a dish, so it shows the mechanism rather than a benefit in a person. In people, a 2022 study shone 810nm LED light on the forearm and measured, through the skin, a rise in oxidised cytochrome c oxidase and in oxygenated haemoglobin. That is the enzyme responding in a living arm. Our ATP research page lists these and the rest wavelength by wavelength.

Why red and near-infrared, and not green

Two reasons. First, the research clusters there: most of the promising results sit roughly in the 600-700nm and 800-900nm bands, which is why panels use the wavelengths they do. Second, the two bands reach different depths. Red is absorbed mostly in the skin and near-infrared penetrates further, and a 2020 review of multi-wavelength photobiomodulation concluded that combining them can present a therapeutic advantage by providing concurrent energy delivery to biological tissues at different depths. Blue is a different question with a different answer, which we wrote up in why our panels have no blue light.

Why not just sit in the sun

Sunlight contains the red and near-infrared bands, and a lot else: ultraviolet, with its well-known costs, and more blue than you want in the evening. The How RLT Works page has the colour-by-colour table. Two more practical problems: clothes cover most of the skin whenever we are outside, and cities are short of near-infrared because built surfaces absorb it while plants and other greenery reflect it. A panel gives you the two useful bands, nothing else, on as much skin as you choose to expose.

More is not better

The dose response is biphasic. Below a threshold nothing happens; above it there is a plateau; push further and the effect falls away again - Huang, Chen, Carroll and Hamblin set this out in 2009 and updated it in 2011. So a longer session is not a stronger session, and a panel that flickers or pulses is delivering less than it looks like it is, which is the point of our flicker test and pulsing posts. For session times, the Dosage Estimator turns your panel and your distance from it into minutes for each goal, and How to do RLT covers the routine.

Is it safe?

For skin, the best data are two phase I dose-escalation trials of red LED light in healthy adults, which found it safe up to 320 J/cm² for skin of colour and 480 J/cm² for non-Hispanic Caucasian participants; the dose-limiting effects at the top of the range were blistering and prolonged redness, and no serious adverse events occurred. A normal panel session is a small fraction of those doses. Eyes are a separate question, covered in our eyesight post. Our devices are not medical devices and are not intended to treat or prevent disease; the disclaimer at the bottom of every post says so for a reason.

Where to go next

If you want to know which device fits what you want to do, the device finder asks six questions and answers that. If you want the studies themselves, start with the wavelength research page. And if you want to ask a person, contact us - Helmut answers himself.