When to Do Red Light Therapy: Morning, Night, Workouts

Short version: for skin, nobody has tested time of day, so nobody knows. Around training it matters more, though not in the direction most people are told. And before either question is worth asking, one distinction has to be nailed down, because getting it wrong is the single most common mistake in this category.

Red light therapy and red light in your evening are two different things. One is a short, high output treatment aimed at your skin. The other is low level ambient room lighting reaching your eyes for hours. They share a color and nothing else, and evidence about one does not transfer to the other in either direction.

The conflation that ruins most timing advice

A therapy panel is a treatment. It runs at an irradiance high enough that manufacturers quote a session in minutes, it is pointed at a patch of skin, and it is over quickly. A red bedroom bulb, an amber lamp or a warm night light is an environmental cue: it reaches your eyes rather than a patch of skin, and it is on for hours rather than minutes. Nobody has published a measurement of what a consumer panel actually delivers to the cornea, so the two cannot even be lined up on intensity.

Almost every "red light is good before bed" article on the internet takes research about the second and applies it to the first. That is not a small slip. The two interventions differ in duration, in what tissue receives the light, and in which biological pathway is even plausible.

Here is the load bearing fact. Circadian entrainment in humans runs through the eye. When bright light was applied to the skin behind the knees in a controlled human experiment, it produced no circadian phase resetting at all, overturning an earlier report of extraocular photoreception. Take it with its limits: it was published as a one page technical report, and it rules out phase resetting from light on skin rather than ruling out local tissue effects of light on skin.

With those limits attached it is still the clearest separation in this whole field between a light therapy device and a circadian light intervention. It is contested in this evidence base by a 20 person cohort study reporting that nightly whole body red light raised serum melatonin, but those are different measurements, that study was not randomized and its control group lay under a switched off machine, and it has never been replicated.

So if your goal is to sleep better, the pathway with actual mechanistic support is the light reaching your eyes in the evening. An expert consensus puts the target at a melanopic equivalent daylight illuminance below 10 lux at the eye in the three hours before bed, below 1 lux in the bedroom, and above 250 lux during the day, all measured vertically at eye level.

Read those as expert consensus midpoints, not trial endpoints. They were synthesised from laboratory dose response work in healthy adults; nobody randomized people to the thresholds, and the authors report that individual sensitivity varies more than tenfold. Melanopic EDI is a spectrum weighted number, so you cannot read it off a household lux meter or infer it from a bulb's color temperature.

And a well established mechanism is not the same thing as a proven intervention. Evening red or amber room lighting has never been tested in a randomized trial for objective sleep. The closest randomized proxy, blue light blocking glasses, was null on every actigraphy measured sleep outcome in a pooled sample of about 49 people. So the honest framing is an asymmetry rather than a winner: a strong mechanism at the eye with thin interventional evidence, against no mechanism and no evidence at all for a panel on your skin.

Two more corrections while we are here. Red is weak, not inert: in a small laboratory exposure delivered to the eyes at high photon density, long wavelength red light did measurably suppress nocturnal melatonin, far less than 460nm blue at the same photon density but not zero.

Against that, 630nm red at 40 lux at the cornea produced no measurable suppression, and a preliminary 12 person study at 80 lux found red matching blue at the one hour mark before diverging. Those three do not agree, and intensity is the likely reconciliation. Nobody has published a measurement of the corneal irradiance or melanopic content a consumer panel actually delivers, so the circadian relevance of an evening session cannot be calculated from anything in the literature.

And melatonin sparing is not the same claim as sleep promoting. In the same experiment where 630nm red at 40 lux left melatonin alone, it raised EEG beta power, lowered alpha power and increased heart rate compared with darkness. Attach the scale: 14 participants, authors labelling their own work preliminary, and the one thing a user would actually notice did not move, because self reported sleepiness and performance were not strongly affected. The alerting turned up on instruments, not in what people said they felt. Light can act through a pathway that has nothing to do with the clock.

Time of day for skin: unstudied, so pick the slot you will keep

There is no trial comparing morning and evening sessions for facial skin, hair, or anything else you would buy a panel for. The optimal time of day is listed as an open question for every application in this domain, including by the authors of the 2025 review of transcranial photobiomodulation for sleep and cognition, who name comparing schedules at different times of day as unresolved research.

The only human study anywhere that gave the same red light exposure at two different times of day was in the eye, not the skin. A single 3 minute exposure at 670nm and 8 mW/cm2 given between 8 and 9am significantly improved color contrast thresholds in aging eyes, still measurable a week later, while the identical exposure between 12 and 1pm did not reach significance.

The catch is that the morning arm had 20 people and the afternoon arm had 6. A weakly powered null in a six person group is not evidence that afternoons do nothing, and retinal color contrast is not skin.

So the honest guidance is boring: choose the time you will realistically hit. Not because consistency itself has been shown to work, but because nothing else has either. No schedule has been shown better than any other, and the trial schedules in this field were set by convention or by manufacturer instruction rather than compared against alternatives. There is no number here to optimize. What that means for how often you run sessions is a separate question from what hour it is.

Before or after a workout

The standard consumer framing is "use it afterwards to recover". The pooled evidence that exists does not test that. Almost all of it comes from light applied before exercise as a preconditioning stimulus, which makes post exercise application under-tested rather than tested and found weaker. Those are not the same finding, and the second is the one usually implied.

A 2024 meta-analysis of 34 randomized trials found pre exercise application improved muscle endurance at a standardized mean difference of 0.31, aided strength recovery at 0.24, and lowered creatine kinase.

Two of those numbers deserve immediate deflation. The authors call 0.31 moderate; by conventional thresholds it is small. And creatine kinase is a blood marker of muscle membrane damage, not a measure of performance or of how recovered you feel, so a change in it should never be reported as faster recovery.

A 2025 meta-analysis pooling overlapping trials put strength recovery after damaging exercise at 0.97 at 24 hours and 0.99 at 48 hours, four times higher on the same construct. That two analyses of largely the same literature land four times apart is the damning part. The larger figure rests on four small studies with 76 to 81 percent heterogeneity and a 48 hour confidence interval whose lower bound is 0.05, meaning the data are equally compatible with almost nothing. The smaller, duller figure is the one to quote.

Then there is the question of who any of it worked in. That same 34 trial meta-analysis reported benefit in sedentary people and in athletes, but no significant benefit in the physically active people in between, which is precisely the person buying a panel for training. No proposed mechanism predicts that pattern, and it came from an unprespecified subgroup analysis of a small trial pool.

The individual trials that tested trained people directly are null. A triple blind placebo controlled trial in 16 judo athletes found no difference from placebo in jump performance, muscle soreness, perceived fatigue or ultrasound echo intensity at any point out to 48 hours. Adding light to six weeks of sprint and explosive squat training in already trained men produced no benefit over placebo, in a trial from an author group that usually publishes positive results. Two crossover trials in 60 physical education students were null on maximal performance and on recovery. And a meta-analysis of 12 trials of running performance found nothing, at a standardized mean difference of 0.13 and p = 0.11, with zero heterogeneity.

Now the part the pre exercise advice usually omits. That "before beats after" framing is an indirect comparison, assembled from separate trials that mostly tested one timing each. It has never been demonstrated head to head at scale.

The one study that did put the timings against each other within the same people is a randomized crossover in 18 rugby athletes comparing no light, before, during and after exercise. It found no differences between conditions in creatine kinase, lactate, or overall functional test performance. A few isolated sub tests came out significant, which is what you expect from four conditions and many outcomes. A double blind crossover in 13 untrained men applying light either 30 minutes or 6 hours before cycling similarly found nothing at either timing, on time to exhaustion, blood flow, plasma nitrite, lactate or anything else measured.

The defensible position: pre exercise is where the pooled evidence happens to sit, post exercise has barely been tested, the one direct comparison found no difference, and if you already train regularly the trials run in people like you came back null.

How far before? Wider than you think

People assume "before training" means minutes before. The trials did not.

The one dedicated timing trial randomized 60 untrained men, 10 per group, to treatment before a damaging quadriceps protocol at 180 joules per muscle group across three sites. The groups treated 5 minutes, 3 hours and 6 hours beforehand all recovered maximum voluntary contraction better at 24 and 48 hours than the group treated 24 hours beforehand, and all three reduced soreness against placebo. The authors describe an effective window of 5 minutes to 6 hours.

Take that with its problems attached, because they are serious. The device combined a 905nm laser, 633nm and 850nm LEDs and a 110 mT static magnetic field, so light on its own was never isolated. The lead author discloses research support from the manufacturer. Ten per group is small for a comparison across this many arms. And it contradicts the cycling trial above, which found nothing at 6 hours before.

The position this evidence base takes on both trials is that neither can support a recommendation about when to use a home panel, and that is worth saying plainly rather than working around. What the pair does rule out is a narrow pre workout window. Nothing in the evidence supports rushing from the panel to the gym, and there is no minute count to hit. Fit the session into your day.

The format problem sitting underneath the timing question

Before optimizing the hour, check the delivery method, because this is where the exercise evidence quietly falls apart. Nearly every positive result above was produced by a laser or LED cluster held in contact with the skin over a specific muscle belly. That is not a panel on a stand, and no timing finding from a contact cluster has been shown to carry across to one.

How the formats compare on evidence generally is worked through in the evidence guide. If you are still choosing hardware, the device comparison covers panels, masks and handhelds properly.

Before bed: the honest answer

No study has measured objective sleep after an evening session with a therapy panel. Not polysomnography, not sleep architecture, not actigraphy. What does exist used other formats, and it points in two directions.

The only sham controlled randomized trial of red and near infrared light worn before bed used a 660, 740, 810 and 870nm cervical collar for 25 minutes every other night for three weeks in 30 people, and found no difference from sham on objective actigraphy. Improvements were self reported and within group only, and the sham arm improved too. The authors themselves describe the trial as underpowered, it was funded by a grant from the device company, the sample was mostly White midlife women, and participants wore an Oura ring that may itself have changed behavior. An underpowered null from a manufacturer funded trial is not a demonstrated absence of effect.

Against that, a 2026 meta-analysis of five sham controlled trials in 240 participants found a 1.25 point improvement on the Pittsburgh Sleep Quality Index, a self reported questionnaire running 0 to 21, with a confidence interval that nearly touches zero. Those five trials mix devices, wavelengths, body sites and populations, transcranial delivery included, which is why only one of them qualifies as a pre bed worn red and near infrared device. Objective null, subjective small positive. Both have to be stated.

Sleep quality has been reported for whole body exposure too, in a fashion. In the only systematic review of that format, none of the five whole body studies showed a benefit to performance or fatigue biomarkers, but two of them reported improved sleep quality. That is a subjective outcome, in a visible light intervention where expectancy cannot be controlled, and it is the whole of the whole body sleep literature.

The frequently quoted melatonin numbers, a rise from 22.2 to 38.8 pg per mL against 21.7 to 23.8 in controls, come from that 20 person cohort study of 658nm whole body light at 30 J/cm2, 30 minutes nightly for 14 nights. Cohort, not randomized. Control group under an unlit device. Sleep measured only by questionnaire. The population was ten elite female athletes per arm, not twenty ordinary adults, and the study's own endurance outcome was not significant. No mechanism by which light on skin raises circulating melatonin has been established, and it sits against the phase resetting null.

Practically: a bright panel in your visual field late at night is, at minimum, a light source. Red light at the eye has produced alerting changes on EEG and heart rate in one small preliminary experiment, though the 14 people in it did not report feeling less sleepy. If you use it in the evening and sleep worse, move it earlier. Nothing that has been measured speaks to an evening panel session specifically, so your own experience is the best signal available to you.

What to actually do

Goal	When	How strong is the basis
Skin	Any time you will keep to	No timing study exists, and no schedule has been shown better than another either.
Hair	Any time you will keep to	Trial schedules were set by convention, never compared by time of day.
Around training	Before, anywhere from minutes to a few hours out, if you do it at all	Pooled indirect evidence sits on before. The one direct within subject comparison found no difference, and trials in trained people are null.
Sleep	Not the tool for this. The pathway with a mechanism is light at your eyes, not on your skin.	Objective sleep after an evening panel session has never been measured, and the nearest randomized test of an evening light intervention was itself null.

Session length and total dose are a separate question, worked through in how long a session should be, and the whole four variable frame lives in the protocol hub. If you want to know when to expect anything to show up, that is the results timeline.

Safety, at any hour

Timing does not change who should avoid this. Photosensitising medication is the main one. Most drug induced photosensitivity is driven by ultraviolet A at 315 to 400nm, well outside what a red panel emits, but the same source names visible light at 400 to 740nm as the action spectrum for some drugs, and that range covers everything a red panel puts out. It is not a clearance to wave anyone through.

The named cases are specific. Porfimer sodium is activated at 630 plus or minus 3nm, and its label requires patients to avoid direct sunlight and bright indoor light for at least 30 days, with some patients remaining photosensitive for 90 days or more. Verteporfin is activated at 689 plus or minus 3nm, with a label that explicitly names prolonged exposure to light emitting medical devices among the sources to avoid for five days after infusion. Protoporphyrin retains an absorption band at 629nm, far weaker than its main 402nm peak and measured in solvent, which shifts in tissue. Individual action spectra vary, and multi wavelength masks often add blue LEDs around 415nm.

Light sensitive conditions are mostly ultraviolet driven, but solar urticaria and cutaneous lupus have documented visible light reactivity and no study has isolated red wavelengths in those patients. Pregnancy is listed as a contraindication by manufacturers as a default, not because anything was found: no study of consumer LED panels or whole body red light in pregnancy exists.

Eye protection. Do not look into the array. Whether goggles are strictly necessary for visible red at consumer irradiances has never been established, which is a reason for caution rather than against it. Near infrared deserves more care than visible red, because invisible output triggers no aversion response and pupil constriction cannot be assumed.

The international exposure limit for the cornea and lens in the 780 to 3000nm band is 10 mW/cm2 for exposures of 1000 seconds or longer. Note what that is: a limit on radiation arriving at the eye, not a limit on what a panel may emit, and not something you can compute from a marketing figure measured at the panel face.

One 2025 clearance summary, for a 633nm and 830nm LED device, lists IEC 62471 for photobiological safety of lamps and lamp systems and IEC 60601-2-57 for non laser light source equipment in its test battery. That shows these standards are in use in this product category, not that every cleared device is tested identically. A panel sold under the general wellness policy is required to meet neither, and inclusion under that policy does not establish that a product is safe or effective for its intended use.

General information, not medical advice. If you are pregnant, take photosensitizing medication, have a light sensitive condition, or are unsure, speak to a doctor first.

FAQ

Is red light therapy better in the morning or at night?

Nobody knows, because it has never been tested for skin, hair or muscle. The only human study giving the same red exposure at two times of day was a retinal experiment where the morning arm had 20 people and the afternoon arm had 6, which is far too weak to generalise. Choose the slot you will keep to.

Should I do red light therapy before or after a workout?

Before is where the pooled evidence sits, but the support is indirect, and post exercise application is under-tested rather than tested and beaten. A 2024 meta-analysis of 34 trials found benefits for pre exercise application, yet those trials mostly tested one timing each, and the same analysis found no significant benefit in physically active people. The only randomized crossover comparing no light, before, during and after within the same 18 athletes found no difference between any of them, and separate trials in judo athletes and in trained men doing six weeks of sprint and squat training were null.

How long before training should I use it?

Wider than most people assume. The one dedicated timing trial found that 5 minutes, 3 hours and 6 hours before all beat 24 hours before, and the authors describe a 5 minute to 6 hour window. That trial used a device combining a laser, LEDs and a static magnetic field, and a cycling trial found nothing at 6 hours, so treat it as ruling out a narrow pre workout window rather than establishing a good one.

Is red light therapy before bed bad for sleep?

No study has measured objective sleep after an evening session with a therapy panel, so there is no evidence either way. The only sham controlled trial of red and near infrared light worn before bed was null on actigraphy, though its own authors call it underpowered and it was funded by the device manufacturer, while a 2026 meta-analysis found a small 1.25 point improvement on a self reported questionnaire. If evening use leaves you sleeping worse, move it earlier.

Is a red light therapy panel the same as a red evening bulb or night light?

No, and treating them as the same intervention is the most common error in this category. A panel is a short, high output skin treatment. An amber or red bulb is low level ambient light reaching your eyes for hours. Evidence about one does not transfer to the other.

Does red light therapy reset my circadian rhythm?

There is no evidence that it does. Circadian entrainment in humans runs through the eye, and bright light applied to skin behind the knees produced no phase resetting at all. No phase response study of a therapy panel has ever been done, and a search of the US FDA 510(k) database for the product codes these devices are cleared under returns no sleep, insomnia or circadian indication. That is a query of specific product codes in one country's database, and it cannot rule out a clearance held elsewhere.

Frequently asked questions