Red Light Therapy Wavelengths — What the Science Shows

Red light therapy — more precisely called photobiomodulation (PBM) — works because specific wavelengths of light interact with biological chromophores in predictable, measurable ways. Not all red or near-infrared light is equal. The difference between 620nm and 850nm is not merely cosmetic; it determines which tissues absorb the energy, how deeply photons penetrate, and which cellular mechanisms are activated.

This guide covers the full therapeutic spectrum from 620nm to 1100nm, drawing on absorption spectroscopy data, tissue optics research, and clinical trials indexed on PubMed. If you are choosing a device — or trying to understand why your device uses the wavelengths it does — this is the reference you need.

How Wavelengths Drive Photobiomodulation

The fundamental principle is selective absorption. Biological tissue contains molecules called chromophores that absorb photons at specific wavelengths. When the right chromophore absorbs the right wavelength, it triggers a biochemical cascade — increased ATP production, modulated reactive oxygen species (ROS), release of nitric oxide (NO), and downstream effects on gene expression and inflammation.

The primary chromophore in PBM is cytochrome c oxidase (CCO), also known as Complex IV of the mitochondrial electron transport chain. CCO is the terminal enzyme in oxidative phosphorylation — the process by which your cells convert oxygen and nutrients into adenosine triphosphate (ATP), the universal energy currency.

Other chromophores matter too. Water absorbs strongly in the infrared range above 1000nm. Melanin absorbs broadly across the visible spectrum. Haemoglobin (both oxygenated and deoxygenated forms) has distinct absorption peaks. Flavins and porphyrins absorb in the blue-to-red range. But for photobiomodulation at red and near-infrared wavelengths, CCO is the dominant target.

The Absorption Spectrum of Cytochrome c Oxidase

Karu (2008) published the definitive action spectrum for CCO, identifying four primary absorption peaks in the red and near-infrared range:

1. ~620nm — a minor peak, marking the beginning of the therapeutic red window
2. ~660nm — a major peak in the visible red range, corresponding to the reduced (unbound) form of CCO
3. ~810–830nm — a significant peak in the near-infrared, corresponding to the oxidised form of CCO
4. ~850nm — a secondary near-infrared peak, closely related to the 810–830nm absorption band

These peaks are not arbitrary. They correspond to the electronic transitions within the copper centres (CuA and CuB) and the haem groups (haem a and haem a3) of the CCO enzyme. When photons at these wavelengths are absorbed, they dissociate inhibitory nitric oxide from the CCO binding sites, restoring electron flow and oxygen consumption. The result: more efficient mitochondrial respiration, more ATP, and a controlled burst of signalling ROS that activates transcription factors like NF-kB and AP-1 (Hamblin, 2017; de Freitas & Hamblin, 2016).

The absorption spectrum explains why the PBM literature — thousands of studies — clusters around a handful of wavelengths rather than spanning the entire spectrum. Devices built around CCO absorption peaks deliver energy where it can actually be used.

Reference: Karu, T.I. (2008). “Mitochondrial Signaling in Mammalian Cells Activated by Red and Near-IR Radiation.” Photochemistry and Photobiology, 84(5), 1091–1099. PMID: 18651871

The Two Optical Windows

Human tissue is not transparent. Photons travelling through skin, fat, muscle, and bone are scattered and absorbed at every layer. However, tissue optics reveals two spectral regions where penetration is significantly greater — the so-called “optical windows” or “therapeutic windows.”

Optical Window I: Red Light (620–700nm)

In this range, absorption by melanin and haemoglobin is declining while water absorption remains negligible. The result is a window where photons can penetrate through the epidermis and into the superficial dermis — roughly 1–3mm for visible red wavelengths, depending on skin type and anatomical location.

The 620–700nm window is ideal for:

• Skin conditions — acne, wound healing, collagen synthesis, photoageing
• Superficial pain — tendonitis, surface inflammation
• Oral and mucosal tissues — where the target tissue is directly accessible

Within this window, 660nm sits at the CCO absorption peak and has the best-documented evidence base. Shorter wavelengths (620–640nm) are still absorbed by CCO but penetrate slightly less and compete more with melanin absorption. Longer wavelengths (670–700nm) penetrate marginally deeper but move away from the CCO absorption maximum.

Optical Window II: Near-Infrared (700–1100nm)

Above 700nm, light becomes invisible to the human eye but continues to interact with tissue. In this range, haemoglobin absorption drops dramatically, melanin absorption is minimal, and water absorption does not become significant until around 970nm. The result is substantially deeper penetration — centimetres rather than millimetres.

The 700–1100nm window is ideal for:

• Deep tissue targets — joints, bones, deep muscles, organs
• Neurological applications — transcranial PBM for brain tissue
• Systemic inflammatory conditions — where deep penetration is required

Within this window, the 810–850nm range aligns with the near-infrared CCO absorption peak and is by far the most studied. Wavelengths above 900nm begin to encounter increasing water absorption, which limits penetration and shifts the mechanism away from CCO towards photothermal effects.

The Valley Between Windows

The 700–770nm range is sometimes called the “dead zone” — it falls between the two CCO absorption peaks and has relatively poor evidence in the PBM literature. Devices using wavelengths in this range are not necessarily useless, but they are operating away from the established chromophore targets. Some researchers have suggested that wavelengths around 730nm may have distinct biological effects (potentially involving opsins or other non-mitochondrial chromophores), but the evidence remains preliminary.

Why 660nm and 850nm Dominate the Market

Walk into any red light therapy device comparison and you will find the same two numbers repeated endlessly: 660nm and 850nm. This is not marketing coincidence — it reflects genuine photobiology.

660nm sits directly on the primary CCO absorption peak in the visible red range. It has the largest body of clinical evidence of any single PBM wavelength. Avci et al. (2013) identified 660nm as the most frequently used wavelength in dermatological PBM studies. It is also relatively inexpensive to produce with LEDs, which are manufactured at enormous scale for horticultural and medical applications.

850nm aligns with the near-infrared CCO absorption peak and offers deep tissue penetration. Ferraresi et al. (2012) used 850nm in landmark studies on muscle recovery and exercise performance. The wavelength penetrates through skin, subcutaneous fat, and muscle tissue to reach joints, bones, and deep structures that 660nm cannot access.

Together, 660nm + 850nm cover both optical windows and both CCO absorption peaks. A device combining the two wavelengths can address surface-level conditions (skin, superficial wounds) and deep tissue targets (joints, muscles, inflammation) in a single treatment session. This dual-wavelength approach has become the industry standard for full-body panels.

The LED manufacturing factor

There is a pragmatic reason too. High-power LEDs are mass-produced at specific wavelengths dictated by semiconductor materials. Red LEDs at 660nm (aluminium gallium indium phosphide, AlGaInP) and NIR LEDs at 850nm (gallium arsenide, GaAs) are among the most efficient and cost-effective options available. Wavelengths like 670nm or 810nm would also work biologically, but the LEDs may be less efficient, more expensive, or harder to source at high power.

Penetration Depth by Wavelength

Penetration depth is one of the most important — and most misunderstood — concepts in PBM. The numbers cited below represent approximate depths at which a meaningful fraction of photons (typically defined as the depth at which intensity drops to 1/e, or ~37%, of the surface value) reach tissue, based on published tissue optics data (Bashkatov et al., 2005; Salomatina et al., 2006; Kolárová et al., 2010).

Actual penetration varies significantly based on:

• Skin pigmentation — melanin absorbs broadly, reducing penetration in darker skin types
• Anatomical location — the scalp is different from the forearm, which is different from the abdomen
• Blood perfusion — more blood means more haemoglobin absorption
• Fat thickness — adipose tissue scatters light differently from muscle
• Device power — higher irradiance pushes more photons deeper (though the percentage absorbed remains the same)

Approximate penetration depths

Wavelength	Penetration Depth	Primary Tissue Targets
620nm	~0.5–1mm	Epidermis, superficial dermis
630nm	~1–2mm	Dermis, superficial capillaries
650nm	~1.5–2.5mm	Dermis, hair follicle bulge
660nm	~2–3mm	Full-thickness dermis, superficial subcutaneous
670nm	~2.5–3.5mm	Deep dermis, subcutaneous transition
810nm	~3–4cm	Muscle, bone surface, joints
830nm	~3–5cm	Deep muscle, bone, neural tissue
850nm	~4–5cm	Deep joints, muscle bellies, organs
940nm	~1–2cm	Reduced (water absorption increases)
1060nm	~0.5–1cm	Further reduced (significant water absorption)

Note the dramatic jump between visible red (~3mm) and near-infrared (~3–5cm). This is the practical difference between the two optical windows. Red light treats what it can reach — skin, surface wounds, gums. NIR light treats what lies beneath — muscles, joints, bones, the brain.

Reference: Bashkatov, A.N. et al. (2005). “Optical properties of human skin, subcutaneous and mucous tissues in the wavelength range 400 to 2000 nm.” Journal of Physics D: Applied Physics, 38(15), 2543. DOI: 10.1088/0022-3727/38/15/004

Wavelength Comparison Table

The table below summarises the key characteristics of each major wavelength used in red light therapy devices. Evidence strength ratings are based on the volume and quality of published clinical trials.

Wavelength	Type	CCO Absorption	Penetration	Key Applications	Evidence Strength	Common Devices
630nm	Red	Moderate	~1–2mm	Wound healing, PDT, dental	Moderate	Clinical/dental lasers, Celluma
650nm	Red	Moderate-High	~1.5–2.5mm	Hair growth, skin rejuvenation	Moderate	Laser caps, hair growth helmets
660nm	Red	Peak	~2–3mm	Skin, collagen, acne, wound healing, anti-ageing	Very Strong	Joovv, MitoRed, PlatinumLED, most panels
670nm	Red	High	~2.5–3.5mm	Retinal health, skin, mitochondrial function	Growing	Specific clinical devices, some panels
810nm	NIR	High	~3–4cm	Brain (transcranial), nerve repair, deep joints	Strong	Vielight (transcranial), clinical lasers
830nm	NIR	Peak	~3–5cm	Deep tissue, muscle, bone healing, orthodontics	Strong	Thor Laser, clinical devices, Celluma
850nm	NIR	High	~4–5cm	Joints, muscles, inflammation, recovery	Very Strong	Joovv, MitoRed, PlatinumLED, most panels
940nm	NIR	Low	~1–2cm	Circulation, haemoglobin interaction	Limited	Speciality devices, pulse oximeters

Key observations from the table

630nm vs 660nm: Both are visible red, but 660nm has a stronger CCO absorption match and slightly deeper penetration. The majority of modern consumer devices have standardised on 660nm. Clinical lasers and photodynamic therapy (PDT) devices sometimes use 630–635nm for historical reasons — many early HeNe (helium-neon) lasers emitted at 632.8nm, and substantial research was built around that wavelength.

810nm vs 830nm vs 850nm: All three fall within the near-infrared CCO absorption band and have strong evidence. The differences are subtle. 810nm is preferred for transcranial applications (the Vielight Neuro uses 810nm, and several Alzheimer’s and TBI trials used this wavelength). 830nm is used in clinical laser systems like Thor and in the FDA-cleared Celluma device. 850nm has become the consumer panel standard, partly because LED efficiency is excellent at this wavelength and the penetration depth is marginally deeper.

940nm and beyond: Once you pass 900nm, water absorption begins to climb. At 940nm, photons are partially absorbed by water in tissue before reaching deep targets. At 980nm (used in some surgical lasers), water absorption is significant enough that the primary mechanism shifts from photobiomodulation to photothermal heating. Wavelengths above 1000nm (1060nm, 1100nm) penetrate poorly and are rarely used in PBM devices.

How to Choose Wavelengths for Different Conditions

Skin Conditions (Acne, Wrinkles, Wound Healing, Collagen)

Recommended: 660nm (primary), with 630nm as an alternative

Skin conditions are surface-level targets. The epidermis is 0.05–0.1mm thick; the dermis extends to roughly 1–2mm. Fibroblasts (the collagen-producing cells) reside in the dermis. Sebaceous glands sit in the upper dermis. Wound healing occurs at the surface.

660nm penetrates 2–3mm — more than enough to reach these targets. Wunsch & Matuschka (2014) demonstrated significant improvements in skin complexion, collagen density, and roughness using 611–650nm light in a controlled clinical trial (PMID: 24286286). Lee et al. (2007) showed 660nm accelerated wound closure and increased collagen synthesis in animal models (PMID: 17463313).

For acne specifically, blue light (415nm) targets porphyrins in P. acnes bacteria, while red light (630–660nm) reduces inflammation and promotes healing. Many acne devices combine both wavelengths.

Joint Pain and Arthritis

Recommended: 850nm (primary), with 810nm or 830nm as alternatives

Joints are deep structures — the knee joint surface is typically 1–2cm below the skin, and deeper aspects of the hip joint can be 5cm or more beneath the surface. Only near-infrared wavelengths have the penetration to reach these targets.

Hegedus et al. (2009) used 830nm on knee osteoarthritis patients and found significant pain reduction and improved microcirculation (PMID: 19735396). Bjordal et al. (2003) conducted a systematic review and meta-analysis finding that PBM at 820–830nm significantly reduced pain and improved function in chronic joint disorders (PMID: 14522484).

850nm offers marginally deeper penetration than 830nm and is widely available in consumer panels. For accessible joints (fingers, wrists, elbows), 830nm or 850nm will both work well. For deeper joints (hips, shoulders), maximising penetration with 850nm and higher irradiance is advisable.

Muscle Recovery and Athletic Performance

Recommended: 850nm, ideally combined with 660nm

Muscle tissue sits beneath the skin and subcutaneous fat layer — typically 0.5–2cm deep for superficial muscles. Near-infrared wavelengths are necessary for effective delivery.

Ferraresi et al. (2012) published a landmark review showing that PBM at 808–850nm applied before or after exercise significantly reduced creatine kinase levels (a marker of muscle damage), decreased delayed-onset muscle soreness (DOMS), and improved time to peak torque (PMID: 22817596). Leal-Junior et al. (2015) confirmed these findings across multiple studies using 810nm and 850nm (PMID: 25803069).

The combination of 660nm + 850nm is popular for athletic recovery because it addresses both surface-level inflammation (red) and deep muscle damage (NIR) simultaneously.

Hair Growth

Recommended: 650nm or 660nm

Hair follicle bulge stem cells and the dermal papilla sit 1.5–4mm below the scalp surface, depending on the follicle and scalp location. Visible red wavelengths (630–670nm) have the right penetration depth to reach these structures.

The FDA has cleared several low-level laser therapy (LLLT) devices for hair growth, predominantly using 650nm or 655nm (e.g., HairMax LaserComb). Kim et al. (2013) demonstrated that 650nm LED treatment significantly increased hair density and thickness in men and women with androgenetic alopecia (PMID: 24078483). Lanzafame et al. (2014) confirmed similar results in a double-blind, sham-controlled trial using 655nm LEDs (PMID: 24421075).

Brain and Neurological Conditions

Recommended: 810nm

Transcranial photobiomodulation requires photons to pass through the scalp, skull, meninges, and cerebrospinal fluid to reach cortical brain tissue. This demands a wavelength with excellent deep penetration and minimal absorption by overlying structures.

810nm has emerged as the standard for transcranial PBM. Naeser et al. (2014) used 810nm LEDs in traumatic brain injury patients and reported improvements in cognitive function (PMID: 25549556). Saltmarche et al. (2017) applied 810nm transcranial PBM in Alzheimer’s disease patients and observed improvements in cognitive scores (PMID: 28211768). The Vielight Neuro, the most widely studied transcranial PBM device, uses 810nm.

Why 810nm and not 850nm? Although 850nm penetrates deeper in soft tissue, the skull’s optical properties at 810nm appear to offer a slightly better transmission window. Tedord et al. (2015) measured photon transmission through human cadaver skulls and found the 800–830nm range optimal (PMID: 26098775).

Wound Healing and Post-Surgical Recovery

Recommended: 660nm (surface wounds), 850nm (deep surgical sites)

For surface wounds, 660nm delivers energy directly to the wound bed where fibroblasts, keratinocytes, and immune cells are actively involved in repair. Gupta et al. (2014) showed 660nm significantly accelerated wound healing in diabetic animal models through increased collagen deposition and angiogenesis (PMID: 25403591).

For post-surgical recovery involving deeper structures (joint replacements, abdominal surgery), 850nm provides the penetration needed to reduce deep tissue inflammation and promote healing of internal structures. Combination therapy (660nm + 850nm) is often recommended for surgical recovery to address both the surface incision and the underlying tissue simultaneously.

Red Light vs Near-Infrared: A Direct Comparison

Feature	Red (620–700nm)	Near-Infrared (700–1100nm)
Visibility	Visible (bright red glow)	Invisible (slight warmth sensation)
Penetration	1–3mm (skin deep)	1–5cm (tissue deep)
Primary chromophore	CCO (reduced form)	CCO (oxidised form)
Best for	Skin, surface wounds, collagen, acne	Joints, muscles, brain, deep inflammation
Warming effect	Minimal	Mild (absorbed by water at longer wavelengths)
Eye safety	Visible — easier to manage exposure	Invisible — greater risk of unnoticed overexposure
LED efficiency	Very high at 660nm	Very high at 850nm
Clinical evidence	Extensive (thousands of studies)	Extensive (thousands of studies)

Do you need both?

For general health and wellness use, the answer is almost always yes. The combination of 660nm + 850nm has become the default for good reason — it covers the broadest range of conditions and tissue depths. However, if your use case is specific:

• Skin-only use (anti-ageing, acne, collagen): A 660nm-only device or a red-dominant panel is sufficient. You do not need to pay extra for NIR if you will never treat deep structures.
• Deep tissue only (knee arthritis, muscle recovery, brain): A device with strong 850nm output is essential. 660nm adds value for surface inflammation but is not strictly necessary for deep targets.
• Dual use (most people): A panel combining 660nm + 850nm in roughly equal proportions gives you the most flexibility.

Individual Wavelength Deep Dives

Each wavelength used in red light therapy has its own characteristics, evidence base, and optimal applications. Detailed guides are available for each:

• 630nm Red Light — The clinical heritage wavelength. Widely used in early laser studies and photodynamic therapy. Moderate penetration, solid evidence for wound healing.
• 650nm Red Light — The hair growth wavelength. FDA-cleared devices predominantly use 650–655nm. Good penetration to follicle depth.
• 660nm Red Light — The gold standard. Peak CCO absorption, maximum evidence, the default wavelength for consumer panels.
• 670nm Red Light — The emerging wavelength. Recent research on retinal health and mitochondrial rescue. Used by Glen Jeffery’s group at UCL.
• 810nm Near-Infrared — The brain wavelength. Standard for transcranial PBM. Used in TBI and Alzheimer’s research.
• 830nm Near-Infrared — The clinical NIR wavelength. Preferred by Thor Laser and Celluma. Strong orthopaedic evidence.
• 850nm Near-Infrared — The deep penetration standard. Dominant NIR wavelength in consumer panels. Excellent for joints and muscles.
• 940nm Near-Infrared — The circulation wavelength. Interacts with haemoglobin. Limited PBM evidence but used in some multi-wavelength devices.
• Wavelength Comparison Chart — Visual comparison of all wavelengths, their properties, and evidence levels.
• Red vs NIR — Detailed head-to-head comparison to help you choose.

Advanced Concepts: Beyond Single Wavelengths

Multi-wavelength therapy

Some newer devices incorporate three, four, or even five wavelengths. The rationale is to hit multiple chromophore targets simultaneously. For example, a device might combine:

• 630nm — porphyrin activation (anti-bacterial for acne)
• 660nm — CCO peak (collagen, healing)
• 850nm — deep CCO peak (joints, muscles)
• 940nm — haemoglobin interaction (circulation)

The evidence for multi-wavelength superiority over dual-wavelength (660 + 850nm) therapy is still emerging. Some studies suggest additive or synergistic effects, but the clinical data is not yet robust enough to declare multi-wavelength devices categorically superior. The risk with adding more wavelengths is that each wavelength receives a smaller share of the total power budget, potentially reducing the effective dose at each target wavelength.

Pulsed vs continuous wave

Wavelength is only part of the equation. Some devices pulse the light at specific frequencies (e.g., 10 Hz, 40 Hz, 73 Hz). Pulsed delivery at 40 Hz has gained attention for neurological applications following the work of Iaccarino et al. (2016) on gamma oscillation entrainment in Alzheimer’s models (PMID: 27929004). However, pulsed delivery does not change the wavelength — it modulates the temporal pattern of photon delivery. See our frequencies guide for more detail.

The biphasic dose response (Arndt-Schulz curve)

Regardless of wavelength, PBM follows a biphasic dose response: too little light produces no effect, an optimal dose produces maximum benefit, and too much light can actually inhibit cellular function. Huang et al. (2009) demonstrated this principle across multiple wavelengths and cell types (PMID: 19764898).

This means that choosing the right wavelength is necessary but not sufficient. You also need the right dose — measured in joules per square centimetre (J/cm2) — which depends on irradiance, treatment time, and distance from the device. See our irradiance and dosing guide for detailed protocols.

Wavelengths to Be Cautious About

Below 620nm (blue, green, yellow)

Wavelengths below 620nm are not “red light therapy.” Blue light (400–490nm) has specific applications — acne treatment via porphyrin activation, circadian rhythm regulation — but operates through entirely different mechanisms. Green light (500–565nm) and yellow light (565–590nm) have limited PBM evidence and poor penetration. Devices marketing “full spectrum” therapy that include blue or green LEDs alongside red and NIR should be evaluated carefully — the non-red wavelengths may not contribute meaningfully to PBM outcomes.

Above 1000nm (far infrared)

Far-infrared (FIR) devices — including infrared saunas — emit wavelengths above 3000nm (typically 6000–15000nm). These work through fundamentally different mechanisms: thermal heating of tissue rather than photobiomodulation of mitochondria. FIR has its own evidence base for circulation, detoxification support, and pain relief, but it should not be conflated with red light therapy or near-infrared PBM.

The 700–770nm “dead zone”

As noted earlier, this range falls between the two CCO absorption peaks. Some inexpensive devices use LEDs in this range because the components are cheap. The biological effects at these wavelengths are not well characterised, and clinical evidence is sparse. This does not mean they are harmful — just that the evidence supporting their therapeutic use is significantly weaker than for 660nm or 850nm.

Practical Recommendations

For first-time buyers

If you are purchasing your first red light therapy device, a dual-wavelength panel combining 660nm and 850nm is the safest, most versatile choice. This covers the largest range of evidence-backed conditions and provides both surface and deep tissue treatment.

For specific conditions

Match the wavelength to the tissue depth. Skin problems warrant 660nm. Joint problems warrant 850nm. Brain applications warrant 810nm (look for purpose-built transcranial devices). Hair loss warrants 650nm. When in doubt, 660nm + 850nm covers most bases.

For advanced users

Consider adding wavelengths to address specific targets: 670nm for retinal and mitochondrial health (based on Jeffery et al., 2021, PMID: 33818515), 810nm for transcranial PBM, or 940nm for enhanced circulation. But ensure your primary wavelengths (660nm and 850nm) are delivering adequate dose — adding wavelengths should not come at the expense of power at the proven wavelengths.

Summary

Red light therapy wavelengths are not interchangeable. The science points clearly to specific wavelength ranges — particularly 660nm and 850nm — where CCO absorption is highest, tissue penetration is optimal, and clinical evidence is strongest. Understanding wavelengths is the first step to getting meaningful results from photobiomodulation therapy.

The key takeaways:

• CCO has two absorption peaks: ~660nm (red) and ~810–850nm (near-infrared)
• Two optical windows allow therapeutic light to reach tissue: 620–700nm (surface) and 700–1100nm (deep)
• 660nm is the gold standard for skin and surface conditions
• 850nm is the standard for deep tissue, joints, and muscles
• 810nm is preferred for transcranial (brain) applications
• Dual-wavelength (660 + 850nm) devices offer the most versatile coverage
• Wavelength selection should match tissue depth — there is no universal “best” wavelength

For detailed analysis of individual wavelengths, explore the deep-dive guides linked above.

---

This article references peer-reviewed studies indexed on PubMed. It is for educational purposes only and does not constitute medical advice. Consult a healthcare professional before beginning any photobiomodulation protocol.