The null case

The worldwide evidence base is 619 people.

The 2023 Cochrane systematic review pooled every randomised trial of blue-light-filtering spectacle lenses it could find: seventeen trials across six countries, with individual trials enrolling between five and 156 people. No trial ran longer than five weeks. None was at low risk of bias across all seven assessed domains. Three of the seventeen clearly masked both participants and staff — which is difficult, because a person can see that their lenses are orange.

How certain the review was, outcome by outcome

GRADE certainty, four levels. More ticks is more confidence in the finding, not a bigger benefit.

• Best-corrected visual acuityModerateProbably little or no effect.
• Adverse eventsLowNo signal of harm, weakly evidenced.
• Visual fatigue scoreLowNo attenuation of eye strain.
• Critical flicker-fusion frequencyLowNo difference detected.
• Sleep qualityVery lowSix trials measured it. Three found a benefit, three found nothing.
• • Serum melatonin
  • Macular health
  • Contrast sensitivity
  • Colour discrimination
  • Discomfort glare
  • Overall visual satisfaction
  No evidenceNot rated low. Not rated very low. Not measured by a single randomised trial in the review.

When sleep was measured by a device instead of a diary

A 2025 meta-analysis pooled the randomised crossover trials that measured sleep with actigraphy rather than a questionnaire. Three trials. Forty-nine adults in total. Every confidence interval crosses zero, and two of the four point estimates point the wrong way.

The authors’ own listed limitations include intervention periods of about a week, healthy participants, no measurement of the light people were actually exposed to, and potential publication bias favouring positive results — which they could not formally test, because fewer than ten studies existed per outcome. That is a field reporting that it is too small to check itself.

The trial that tested the expectation instead of the lens

In 2021 a double-masked randomised trial put 120 symptomatic computer users through a two-by-two design: each person watched a recorded clinician either endorsing the intervention or not, then received blue-blocking or clear placebo lenses, with everyone told they had received something active. Neither the endorsement (p = 0.410) nor the lens type (p = 0.394) changed the eye-strain score. It is the only trial in the field that built the expectation into the design, and it is null on both axes.

The one positive result the category leans on

A 2017 University of Houston study had 22 people wear short-wavelength-blocking glasses for three hours before bed for two weeks and reported roughly 58% higher night-time melatonin, with self-reported improvements in sleep. It is a real, honestly conducted, small study. Twenty-two people. Nobody was blinded to an orange lens. The sleep outcome was self-reported. It has not been replicated at scale, and the device-measured pooled results above are null. We are stating it because leaving it out would be its own kind of dishonesty, not because we are building a claim on it.

What the people who reviewed it said

Commenting on the Cochrane review, Professor Russell Foster of the University of Oxford said the evidence for claims about sleep, eye strain and retinal protection was at best weak, and at worst non-existent. The American Academy of Ophthalmology does not recommend blue-light-blocking glasses and states there is no scientific evidence that light from computer screens damages the eyes.

One thing the review does not settle

The trials it pooled overwhelmingly tested clear or lightly filtering lenses, worn during the day, for eye strain. That is a different product from a deep amber or red lens worn in the evening. The review is decisive against the first. It is close to silent on the second — and silence is not support. Anyone citing Cochrane as proof that evening lenses work, or as proof that they cannot, is reading past the end of the evidence in whichever direction suits them.

What is actually established.

Three things in this area are well supported. All three are statements about light in a room, measured in melanopic units. None of them is a statement about a pair of glasses, and the distance between the two is where this industry lives.

Human non-visual photoreception peaks near 480 nm

Which is why international light recommendations are written in melanopic equivalent daylight illuminance rather than in ordinary lux. It is a fact about a receptor, formalised in a published standard. It is not a fact about a product.

Light at night suppresses melatonin

Not in dispute. What is in dispute — because nobody has measured it in a trial of these lenses — is how far any consumer pair moves that number in a real living room with the ceiling light on.

There are consensus numbers, and they are about rooms

A 2022 international expert consensus recommends a minimum of 250 lux melanopic EDI at the eye during the day, a maximum of 10 lux in the three hours before bed, and a maximum of 1 lux while asleep. The authors attach their own caveats: more than tenfold variation between individuals, derived for healthy adults aged 18 to 55, explicitly not intended for shift workers. We can point you at the numbers. We cannot tell you what any product does to them in your home, and neither can anyone else.

The five places the marketing jumps the gap

Every claim in this category that fails does so by crossing one of these five gaps in a single sentence, usually without noticing.

1. A receptor→an outcome

   Human non-visual photoreception peaks near 480 nm. That is established receptor biology. It does not license the sentence that follows it in almost every advertisement, which is that filtering 480 nm therefore improves anybody's sleep. This is the specific move the UK advertising regulator banned a lens brand for in 2022.

2. A mechanism→a product

   Light at night suppresses melatonin; that is not in dispute. “Our lenses protect your melatonin” is a different statement about a specific object, and not one randomised trial in the Cochrane review measured melatonin at all.

3. A percentage→an effect size

   How much light a lens transmits is an optical fact about a piece of plastic. How much light reaches a retina in a real room is a different quantity, involving the light arriving around the frame, the rest of the room, and — as an unresolved question in the literature — the pupil widening behind a darker lens. A blocking percentage is not a dose, and a dose is not an outcome.

4. A laboratory→a bedroom

   Acute melatonin-suppression experiments run under controlled dim light in a lab are not measurements of how anyone sleeps at home. The published international recommendations built on that work are stated as light levels in a room, not as eyewear.

5. Patients→everybody

   The 2025 meta-analysis restricts its cautious optimism to specific clinical subgroups rather than the general population. Borrowing a clinical result and printing it on a consumer product is the oldest move in this category.

Claims we will not make.

This list costs us conversions. It is here because every sentence on it is a sentence somebody in this category is currently printing, and because a claim you cannot substantiate is a claim you are one complaint away from deleting.

• That these lenses improve sleep — in any wording. “May help you sleep”, “supports better sleep” and “designed to improve sleep” are the same claim wearing a coat.
• That they protect, preserve or raise melatonin. No randomised trial of these lenses has measured melatonin.
• “Clinically proven”, “scientifically proven”, “backed by science”, “science-based”.
• That they reduce eye strain, fatigue or headaches. The best-designed trial in the field tested exactly this and found nothing.
• That they protect your retina, prevent macular degeneration, or block “harmful” blue light. Two brands have been banned for this wording.
• “Doctor recommended”, “optometrist approved”, “recommended by sleep scientists”.
• A percentage without the wavelength band, the method and the report behind it.
• Any before-and-after sleep chart, sleep-score projection, or with-and-without comparison of a person.
• “FDA approved”. There is no such thing in this category.

And what happened to the brands that did

Boots Opticians, UK

Advertising ruling upheld in 2015; the General Optical Council fined the company £40,000 in 2017 and the lenses were withdrawn. The graph Boots supplied showed 20% of blue light filtered, and the company could not connect that percentage to any reduction in harm. A technically accurate chart was itself the misleading element. Source

Gamer Advantage, US brand advertising in the UK

Ruling upheld in July 2022 over claims that the lenses were clinically proven to stop melatonin suppression for a better night's rest. The regulator found the submitted evidence — an unpublished thesis, unpublished posters and a news article — inadequate, and told the company not to claim its glasses could improve sleep quality without adequate substantiation. Source

Reticare

Ruling upheld in May 2024. A rat study and two patents describing cell cultures under twelve hours of continuous exposure were rejected as not representing consumer screen use. Source

Gunnar Optiks

A US class action filed in April 2023 alleges the company's published blue-light blocking figures fall short of what the lenses achieve. These are allegations in a filed complaint. We have not retrieved a docket or a disposition and make no statement about the outcome. Source

How to check any lens, including ours.

• The blue-light pen in the box is not a test. Those pen LEDs peak near 400 nm, in the violet. Screens emit their blue nearer 450–460 nm. A lens can extinguish the pen completely and still pass most of a screen’s blue. A brand that ships one as proof is either not measuring or not telling.
• Ask for a curve, not a number. A spectral transmittance curve from 380 to 780 nm, from a named laboratory, for the exact lens being sold. Three of the nine products on our comparison page publish anything of the kind.
• Read the band on every percentage. 400–500 nm is the flattering band. The published survey of 50 commercial lenses weights across 455–560 nm, which includes the green half that most marketing stops before.
• Distrust two phrases in particular. “Up to X%” is a peak, not an average. “N times more than a standard lens” is a multiple of a baseline nobody has told you.

Sources

• Cochrane review, 2023 — blue-light filtering spectacle lenses (CD013244.pub2)
• Singh, Downie & Anderson, 2021 — double-masked randomised trial, Am J Ophthalmol
• Luna-Rangel et al., 2025 — actigraphic sleep outcomes, meta-analysis, Front Neurol
• Mason et al., 2022 — spectrophotometric properties of 50 commercial blue blockers, Chronobiol Int
• Brown et al., 2022 — recommendations for indoor light exposure, PLOS Biology
• Ostrin et al., 2017 — University of Houston College of Optometry release
• Science Media Centre — expert reaction to the Cochrane review
• American Academy of Ophthalmology — are computer glasses worth it?

Nothing on this page is medical advice or a diagnosis. If evening light, screen use or sleep is a problem for you, that is a conversation for a clinician, not for a shop.

See the lens comparison →