Research synthesis · 28 September 2026
Screen flicker and digital eye strain: what the research shows, and what it does not
Eye and vision symptoms from screen use are common enough to have a name: digital eye strain, also called computer vision syndrome. Estimates suggest it may affect half or more of computer users.1
The popular explanation blames blue light. This page reads six bodies of research side by side: blinking, flicker, display dimming, evening light, brightness and contrast, and individual sensitivity. For each we say what was measured, in whom, and what each abstract does not report.
Our short reading: in the studies below, eye strain symptoms were linked to incomplete blinking, breaks and glare, and the tested blue-light lenses showed no clear short-term benefit for eye strain. The evidence that flicker you cannot see causes eye strain is thin, and most of it comes from lighting and laboratory stimuli rather than screens. We say so wherever it applies.
1 · The eye surface
Blinking, incomplete blinks and screen breaks
Reviews sort the symptoms into two families: strain on focusing and on keeping the two eyes aligned, and external symptoms linked to dry eye. The symptoms are usually transient, but they can be frequent and persistent.1
When 21 people read on a desktop computer for 15 minutes, those who made more incomplete blinks reported more symptoms. Roughly doubling their blink rate with an audible tone did not produce a significant change in their symptom scores.2
When a computer screen was compared with printed text matched for size, contrast and viewing distance, blink rates did not differ significantly, but incomplete blinks were significantly more common on the screen. The authors proposed that earlier reports of reduced blinking reflect the mental demand of the task rather than the screen.3
A study of 50 readers found the same shape. Every reading condition, on paper or on a screen, lowered blink rate against a baseline of looking at a landscape picture, but only reading on a screen raised the share of incomplete blinks.4
In 29 symptomatic computer users, two weeks of software reminders to follow the 20-20-20 rule reduced eye strain and dry eye symptoms, but the improvement was not maintained a week after the reminders stopped, and no change was observed in any tear film or eye surface parameter.5
2 · The evidence against the popular story
Trials of blue-light filtering lenses for eye strain
A 2023 Cochrane systematic review examined 17 randomised trials of blue-light filtering spectacle lenses. It concluded that these lenses may not reduce symptoms of eye strain from computer use over short follow-up, compared with ordinary lenses.6
The review found no clinically meaningful difference in critical flicker-fusion frequency, one of the measures used as an index of visual fatigue, and it rated the evidence on both outcomes as low certainty.1,6 Effects on sleep quality were indeterminate, and the review states that none of the included studies evaluated serum melatonin.6
Two smaller studies point the same way. In a double-blind test, 24 people reading from a tablet for 20 minutes reported more symptoms afterwards, with no significant difference between blue-blocking and plain lenses.7 In 23 young adults reading from a computer for 30 minutes, a blue-blocking filter did not significantly change the activity of the muscle around the eye or the reported symptoms.8
3 · Temporal light modulation
Flicker, visual discomfort and the visual cortex
Flickering light can be uncomfortable to look at and can trigger seizures in people with photosensitive epilepsy.9 How uncomfortable depends on the shape of the flicker over time. In lab tests, discomfort rose as the flicker departed from the temporal patterns of natural scenes, and its phase structure mattered as well.9A follow-up found that one waveform’s relative comfort held only at daylight-range (photopic) light levels.10
In 20 people viewing flicker between about 1.6 and 30 Hz, the size of the visual cortex’s response related directly and linearly to how uncomfortable the flicker felt.11
The brain can register flicker a person cannot see. Imaging showed that many visual areas responded differently to colour flicker at 25 Hz or faster, which observers saw as one steady colour, than to a steady control.12 Its abstract reports no discomfort or strain measure, so on its own it shows a cortical response, not harm.
Flicker too fast to notice can reappear when the eyes move. A 2026 narrative review, whose first author is listed as affiliated with Grena Limited, London, notes that such modulation may cause visual discomfort during rapid eye movements through the phantom array effect, and in the lab a flickering light seen during a saccade breaks into a sequence of dashes whose colours different people report differently.14,15 In a reading study under LED light, reading was slowest under 60 Hz and 600 Hz flicker, the latter being the rate at which the phantom array is most visible.13
The same review reports that foundational research found continuous exposure to 100 Hz lighting can cause headaches and eyestrain even when the flicker is not seen. It also says the strongest evidence concerns visible flicker in photosensitive epilepsy, that direct evidence linking typical LED modulation to seizures is missing, and that the evidence suggests caution, not causality.15
4 · Display dimming
Pulse-width modulation: where flicker meets the dimmed screen
Some OLED smartphone displays dim by pulse-width modulation (PWM), and manufacturers compare them using a flicker-visibility measure that was developed for lighting products.16
One laboratory study presented an oscillating arm in front of an OLED smartphone display to find when the resulting stroboscopic effect became visible. At dimming frequencies of 480 and 960 Hz it was visible when display luminance was below 40 cd/m², while at 1920 or 2160 Hz it was not visible.16
5 · Circadian timing
Evening screen light and the body clock
In a controlled comparison, people who read a light-emitting e-book before bed instead of a printed book took longer to fall asleep, secreted less melatonin, had a later circadian clock and were less alert the next morning.21
Ordinary room light before bedtime, compared with dim light, suppressed melatonin, delayed its onset in nearly all of 116 young adults and shortened the duration of its production.22 That was room lighting, not a screen.
Amount and duration matter. With tablets at their highest brightness, suppression after one hour was not statistically different from zero in 13 participants, but after two hours it was.23
One screen study controlled melanopic irradiance independently of display luminance and colour. Each of 72 healthy men was exposed in the evening, before habitual bedtime, to one of four luminance levels at a low and a high melanopic setting. Time to fall asleep, melatonin concentration and melatonin onset changed with melanopic irradiance in a dose-dependent way; subjective alertness did not.24
Not every study agrees. After a day of bright light, two hours of reading on a tablet or on paper made no difference to sleep or to pre-sleep saliva melatonin in 14 people, and the authors call for larger studies.25 In 13 teenage boys at an LED screen, blue-blocking glasses reduced evening suppression of melatonin compared with clear lenses, but visually scored sleep stages and the next morning's measures were not modified.26
6 · Luminance and contrast
Brightness, contrast and the room around the screen
In a two-hour visual task, a glare environment raised visual fatigue scores and a glare-free display lowered them. The rise in fatigue scores was smaller with uniform supplementary room lighting, and visual performance was worse in dim or uneven lighting.27 Critical fusion frequency was the only visual function parameter that decreased significantly after the task.27
In proofreading experiments, dark text on a light background gave better performance than light text on a dark background, in darkness and in office light alike, while self-reported eyestrain and headache did not vary as a function of any of the conditions.28
In 32 young adults, visual comfort fell for stimuli made to look brighter by an illusion, especially under low ambient light.29
7 · Individual differences
Migraine, light sensitivity and flicker discomfort
Light sensitivity is, after headache, the most debilitating symptom reported by people with migraine.17In 60 people, those with migraine did not differ from headache-free people in how melanopsin and cone signals were combined, but showed an amplified discomfort response to the combined signal from the retina’s melanopsin-containing cells (ipRGCs).17
A second study found that people with migraine, tested between attacks, had lower thresholds for light-induced discomfort, and that melanopsin contributed about one and a half times as much as cone luminance to those thresholds.18 In a study of 14 people with migraine and 14 controls, those with migraine had significantly elevated flicker discomfort scores, which correlated significantly with the number of years they had had migraine.19
The picture is not uniform. In a study of 10 people with migraine with aura and 10 controls, cortical responses to flicker and the discomfort that went with them did not differ substantively between groups.11 A review argues that in migraine a cortical mechanism explains discomfort from light, flicker, patterns and colour more simply than the retinal cells alone.20
Synthesis
Connecting the literatures: our reading, labelled as such
The points below are our interpretation of the studies above. Each one is a link between findings, not a finding of its own.
- Eye strain and the body clock are separate questions. The eye strain evidence points at blinking, breaks and glare.2,3,4,5,27 The evening-light evidence points at the amount, duration and melanopic content of light.23,24 In the lens trials, the tested lenses showed no clear short-term benefit for eye strain.6,7,8 A change aimed at one question should not be assumed to answer the other.
- The same retinal pathway appears in two fields. The 72-person evening-screen study described its light by melanopic irradiance,24 and the migraine studies implicate melanopsin signals in light-induced discomfort.17,18 The sleep abstracts report clock and sleep outcomes and the migraine abstracts report discomfort. We found no abstract reporting both in the same people at the same screen.
- Dimming may pull in two directions on some hardware. Less light, and less melanopic light, in the evening is the direction the circadian studies point.23,24 On one PWM-dimmed phone display, low luminance at lower dimming frequencies is also where the stroboscopic effect became visible.16 This is a question to test, not a finding.
- Flicker sits between the two stories. It is a documented source of discomfort in laboratory flicker studies,9,10,11 a measurable brain signal when invisible,12 and an open question as a cause of everyday screen strain.15
Open questions
What nobody has measured yet, as far as we could find
Each item below is a gap we found in our PubMed search, not a claim that no such work exists anywhere.
- Whether PWM dimming at the frequencies used in laptops and monitors changes eye strain or headache, measured as symptoms in masked conditions. The abstract of the display study we found reports visibility on one smartphone display only.16
- Dose-response studies relating modulation depth to headache, eyestrain, visual discomfort and task performance, which the 2026 narrative review names as the priority.15
- Whether people with migraine or high visual sensitivity respond differently to display PWM than other people do.
- Whether changing the colour of the display itself, rather than wearing a filter, changes eye strain symptoms in a masked trial.
- Whether evening screen settings that lower melanopic light also change blinking or the eye surface.
- Whether daytime light exposure changes how much evening screen light matters, which one small study suggested and asked larger studies to test.25
Reusable
The evidence table
One row per study or pair of studies, restating only what each abstract reports. You may reproduce this table with a link back to this page.
| What was measured | In whom | Exposure | Result | Limits | Source |
|---|---|---|---|---|---|
| Blink rate, incomplete blinks, symptom score | 21 subjects | 15-minute reading on a desktop computer | More incomplete blinks went with more symptoms; raising blink rate produced no significant change in symptoms | Small; short laboratory tasks; correlation | Portello 2013 |
| Blink rate and incomplete blinks, screen vs paper | 25 subjects | 20-minute reading, matched text on screen or paper | No significant blink-rate difference; significantly more incomplete blinks on screen | Small; short laboratory tasks | Chu 2014 |
| Eye strain and dry eye symptoms and signs | 29 symptomatic computer users | Two weeks of 20-20-20 break reminders | Symptoms fell; the gain was not maintained a week after stopping; no change observed in any ocular surface or tear film parameter | The abstract reports no control group | Talens-Estarelles 2023 |
| Eye strain, flicker-fusion frequency, sleep quality | 17 randomised trials of adults | Blue-light filtering spectacle lenses | May not reduce eye strain; no clinically meaningful fusion-frequency difference; sleep effects indeterminate | Low-certainty evidence; the review states no included study evaluated serum melatonin | Singh 2023 |
| Eye strain symptoms; eyelid muscle activity | 24 subjects; 23 young adults | Tablet or computer reading with or without a blue-blocking filter | No significant symptom difference (tablet study); filter did not significantly alter muscle activity or symptoms (computer study) | Short laboratory tasks; small samples | Rosenfield 2020, Vera 2023 |
| Rated discomfort from flicker | Lab participants | Uniform flickering fields with varied waveforms and light levels | Discomfort depends on the flicker's temporal spectrum, phase and light level | Laboratory flicker stimuli; not an ordinary screen-use task | Yoshimoto 2017, Yoshimoto 2020 |
| Visual cortex response and discomfort | 20 people, half with migraine with aura | Flicker from about 1.6 to 30 Hz | Larger cortical response went with more discomfort; no substantive group differences | Laboratory flicker stimuli; small groups | Patterson Gentile 2020 |
| fMRI response to colour flicker | Human observers | Colour flicker at 25 Hz or faster, seen as steady | Many visual areas still responded to the flicker | The abstract reports no discomfort or strain measure | Jiang 2007 |
| Reading speed | Participants grouped by pattern-glare sensitivity | LED light flickering at 60 Hz, 120 Hz or 600 Hz vs steady | Slowest reading at 60 Hz and 600 Hz | LED lighting on printed text; the abstract reports no screen condition | Laycox 2024 |
| Visibility of the stroboscopic effect | Observers (number not given in the abstract) | Oscillating arm viewed in front of an OLED smartphone display dimmed by PWM at 480 to 2160 Hz | Visible at 480 and 960 Hz below 40 cd/m²; not visible at 1920 or 2160 Hz | The abstract reports visibility only and no symptom outcome; one smartphone display | Guo 2025 |
| Light-induced discomfort and pupil response | People with migraine between attacks and controls | Lights targeting melanopsin and cones | Migraine amplified discomfort; melanopsin weighed more than cone luminance | Laboratory light stimuli; not an ordinary screen-use task | McAdams 2020, Zele 2021 |
| Sleep onset, melatonin, circadian timing, alertness | Participants (details in the full paper) | Light-emitting eBook vs printed book before bed | Later sleep onset and clock, less melatonin, less morning alertness | The abstract reports no eye strain outcome | Chang 2015 |
| Sleep latency, melatonin, alertness | 72 healthy men | Screen-level light with low or high melanopic content, in the evening | Dose-dependent effects on sleep latency and melatonin; not on subjective alertness | Men only; the abstract reports no eye strain outcome | Schöllhorn 2023 |
| Sleep and saliva melatonin | 14 participants after bright daytime light | Two hours of tablet vs paper reading | No differences in sleep parameters or pre-sleep saliva melatonin | Small; authors ask for larger studies | Rångtell 2016 |
| Visual fatigue score and visual function | Participants at a two-hour visual task | Glossy, matte or glare-free display; dim or supplementary room light | Fatigue scores higher with glare, lower with a glare-free display; smaller rise with uniform supplementary lighting | Lighting and display surface varied; the abstract reports no display-colour condition | Lin 2019 |
Citable figures
Numbers with their sources
Each line states one figure the way the study’s abstract reports it, with the number, the people and the condition, and the abstract’s own words underneath so you can check it. Every figure comes from the abstract of one of the 29studies cited on this page. Where an abstract reports another study’s number, the line says so, and you should cite the original. The abstracts were first read on 27 September 2026 and read again for this section on 30 September 2026.
Most of these are small laboratory studies, so a figure describes those people under those conditions and not everyone at every screen. The evidence that flicker you cannot see causes eye strain is thin: the flicker figures below come from lighting, one phone display and laboratory stimuli, not from ordinary computer use. If you quote a figure, please cite the study itself, and link this page if it helped you find it.
Eye strain symptoms and blinking
- 01
A 2018 review states that estimates suggest digital eye strain may affect 50% or more of computer users.
“estimates suggest its prevalence may be 50% or more among computer users”
Sheppard AL, Wolffsohn JS (2018), BMJ Open Ophthalmology. PMID 29963645 ↗
- 02
In 21 people reading on a desktop computer for 15 minutes, the mean blink rate was 11.6 blinks per minute.
“Subjects (N = 21) were required to perform a continuous 15-minute reading task on a desktop computer”
“The mean blink rate during the task without the audible tone was 11.6 blinks per minute (SD, 7.84).”
Portello JK, Rosenfield M, Chu CA (2013), Optometry and Vision Science. PMID 23538437 ↗
- 03
In the same 21 people, the share of blinks that were incomplete ranged from 0.9% to 56.5% per person, with a mean of 16.1%.
“The percentage of blinks deemed incomplete for each subject ranged from 0.9 to 56.5%, with a mean of 16.1% (SD, 15.7).”
Portello JK, Rosenfield M, Chu CA (2013), Optometry and Vision Science. PMID 23538437 ↗
- 04
In the same 21 people, total symptom scores showed a significant positive correlation with the percentage of incomplete blinks during the task (p = 0.002).
“A significant positive correlation was observed between the total symptom score and the percentage of incomplete blinks during the task (p = 0.002).”
Portello JK, Rosenfield M, Chu CA (2013), Optometry and Vision Science. PMID 23538437 ↗
- 05
In a second session with the same people, raising their mean blink rate to 23.5 blinks per minute with an audible tone did not produce a significant change in their symptom scores.
“In a second session, the blink rate was increased by means of an audible tone that sounded every 4 seconds”
“Increasing the mean blink rate to 23.5 blinks per minute by means of the audible tone did not produce a significant change in the symptom score.”
Portello JK, Rosenfield M, Chu CA (2013), Optometry and Vision Science. PMID 23538437 ↗
- 06
In 25 people reading the same text for 20 minutes, mean blink rates were 14.9 per minute on a computer screen and 13.6 on paper, not a significant difference (p = 0.58).
“Subjects (N = 25) were required to perform a continuous 20-minute reading task from either a desktop computer screen or a printed hard copy page”
“Identical text was used in the two sessions, which was matched for size and contrast.”
“Mean blink rates for the computer and hard copy conditions were 14.9 and 13.6 blinks per minute, respectively (p = 0.58).”
Chu CA, Rosenfield M, Portello JK (2014), Optometry and Vision Science. PMID 24413278 ↗
- 07
In the same study, incomplete blinks made up 7.02% of blinks on the screen and 4.33% on paper, a significant difference (p = 0.02), but in this study the share of incomplete blinks did not correlate significantly with symptom scores.
“a significantly higher percentage of incomplete blinks was observed for the computer condition (7.02 vs. 4.33%; p = 0.02)”
“No significant correlation was found between the symptom score and the percentage of incomplete blinks.”
Chu CA, Rosenfield M, Portello JK (2014), Optometry and Vision Science. PMID 24413278 ↗
- 08
In 50 participants aged 18 to 74, all six 6-minute reading conditions, on paper or on a screen, lowered spontaneous blink rate against a baseline of viewing a landscape picture (all P < 0.001).
“A sample of 50 participants (29 females, age range, 18-74 years) were recruited for this study.”
“observed in silence a landscape picture at 2 m (baseline) and during six different, 6-minute controlled reading experimental conditions”
“Texts were presented in electronic (tablet and computer display at 100% and 330% zoom levels) and hard-copy (text in book position in silence and aloud and text pasted on the computer display) formats.”
“All reading conditions resulted in a decrease in SEBR when compared with baseline conditions (all P < 0.001)”
Argilés M, Cardona G, Pérez-Cabré E, Rodríguez M (2015), Investigative Ophthalmology & Visual Science. PMID 26517404 ↗
- 09
In 29 symptomatic computer users measured before and after two weeks of reminders to follow the 20-20-20 rule, digital eye strain and dry eye symptoms decreased (p ≤ 0.045), and the improvement was not maintained one week after the reminders stopped (p > 0.05).
“was downloaded onto the laptops of 29 symptomatic computer users”
“were assessed before and after two weeks of using the reminders and one week after the discontinuation of the strategy”
“Dry eye symptoms and DES decreased with the rule reminders (p ≤ 0.045), although this improvement was not maintained one week after discontinuation (p > 0.05).”
Talens-Estarelles C, Cerviño A, García-Lázaro S, Fogelton A, Sheppard A, Wolffsohn JS (2023), Contact Lens and Anterior Eye. PMID 35963776 ↗
- 10
In the same study, no changes in any ocular surface or tear film parameter were observed with the reminders (p ≥ 0.089).
“No changes on any ocular surface and tear film parameter were observed with the rule reminders (p ≥ 0.089).”
Talens-Estarelles C, Cerviño A, García-Lázaro S, Fogelton A, Sheppard A, Wolffsohn JS (2023), Contact Lens and Anterior Eye. PMID 35963776 ↗
- 11
In a study of reflected glare and visual field lighting during a two-hour visual task, critical fusion frequency was the only visual function parameter that decreased significantly after the task, and the authors call it an effective indicator of computer vision syndrome.
“The aim of this study is to investigate the effects of reflected glare and visual field lighting on computer vision syndrome.”
“participants performed a two-hour visual task”
“Critical fusion frequency was the only visual function parameter which decreased significantly after the visual task.”
“Critical fusion frequency is an effective indicator of computer vision syndrome.”
Lin CW, Yeh FM, Wu BW, Yang CH (2019), Clinical and Experimental Optometry. PMID 30805993 ↗
- 12
In 32 young adults with healthy vision, visual comfort decreased for stimuli made to look brighter by an illusion, especially under low light.
“Thirty-two young adults with healthy vision”
“visual comfort decreased for bright illusions, especially under low light”
Rodán A, Fernández-López A, Vera J, Montoro PR, Redondo B, Prieto A (2025), Vision (Basel). PMID 41133605 ↗
Blue-light lenses and filters
- 13
The review included 17 randomised trials, with 5 to 156 participants each and follow-up from less than one day to five weeks.
“We included 17 RCTs, with sample sizes ranging from five to 156 participants, and intervention follow-up periods from less than one day to five weeks.”
Singh S, Keller PR, Busija L, et al. (Cochrane Review) (2023), Cochrane Database of Systematic Reviews. PMID 37593770 ↗
- 14
The reviewers judged 65% of the included studies at high risk of detection bias, because outcome assessors were not masked, and 59% at high risk of performance bias, because participants and personnel were not masked.
“We judged 65% of studies to have a high risk of bias due to outcome assessors not being masked (detection bias) and 59% to be at high risk of bias of performance bias as participants and personnel were not masked.”
Singh S, Keller PR, Busija L, et al. (Cochrane Review) (2023), Cochrane Database of Systematic Reviews. PMID 37593770 ↗
- 15
In that review, 35% of the included studies were pre-registered on a trial registry.
“Thirty-five per cent of studies were pre-registered on a trial registry.”
Singh S, Keller PR, Busija L, et al. (Cochrane Review) (2023), Cochrane Database of Systematic Reviews. PMID 37593770 ↗
- 16
One included trial of 120 participants reported no difference in visual fatigue scores between the two arms: the mean difference was 9.76 units in the direction of worse symptoms, with a 95% confidence interval from -33.95 to 53.47.
“One RCT reported no difference between intervention arms (mean difference (MD) 9.76 units (indicating worse symptoms), 95% confidence interval (CI) -33.95 to 53.47; 120 participants).”
Singh S, Keller PR, Busija L, et al. (Cochrane Review) (2023), Cochrane Database of Systematic Reviews. PMID 37593770 ↗
- 17
For sleep, six trials with 148 participants disagreed, three reporting a significant improvement in sleep scores with filtering lenses and three reporting no significant difference, and the review rates this evidence as very low certainty.
“with respect to sleep quality (very low-certainty evidence)”
“Inconsistent findings were evident across six RCTs (148 participants); three studies reported a significant improvement in sleep scores with blue-light filtering lenses compared to non-blue-light filtering lenses, and the other three studies reported no significant difference between intervention arms.”
Singh S, Keller PR, Busija L, et al. (Cochrane Review) (2023), Cochrane Database of Systematic Reviews. PMID 37593770 ↗
- 18
In 24 people reading from a tablet for 20 minutes, symptoms rose significantly after the task (p = 0.00001), with no significant difference between two blue-blocking lenses and a plain control lens (p = 0.74).
“Twenty-four subjects were required to perform a 20-minute reading task from a tablet computer.”
“They wore either lenses containing a blue-blocking filter (TheraBlue 1.67 or TheraBlue polycarbonate) or a CR-39 control lens which did not include a filter.”
“While a significant increase in symptoms was observed immediately following the near vision task (p = 0.00001), no significant difference in symptoms was found between the 3 lens conditions (p = 0.74).”
Rosenfield M, Li RT, Kirsch NT (2020), Work. PMID 32007978 ↗
- 19
In 23 healthy young adults (mean age 22.9 years) reading from a computer screen for 30 minutes, a Bayesian analysis favoured no change in the activity of the muscle around the eye with or without a blue-blocking filter, with a Bayes factor of 7.08 in favour of no change.
“Twenty-three healthy young adults (22.9 ± 3.2 years of age) performed two reading tasks from a computer screen with or without a B-B filter”
“during a 30-min reading task from a computer screen”
“A Bayesian analysis favoured the null hypothesis that there was no change in OO muscle activity with or without using the B-B filter (Bayes Factor01 [BF01] = 7.08).”
Vera J, Redondo B, Ortega-Sanchez A, et al. (2023), Clinical and Experimental Optometry. PMID 35057697 ↗
Flicker detection and visibility
- 20
In two experiments with text lit by LED lights flickering at 60 Hz, 120 Hz or 600 Hz, compared with 60 kHz flicker perceived as steady light or with continuous light, reading speed was slowest during the 60 Hz and 600 Hz flicker.
“We measured reading speed while striped and less striped texts were illuminated with LED lights.”
“In Experiment 1, the lights flickered at 60 Hz and 120 Hz compared to 60 kHz (perceived as steady light).”
“In Experiment 2, the lights flickered at 60 Hz or 600 Hz (at which frequency the phantom array is most visible), and were compared to continuous light.”
“Overall, reading speed was slowest during the 60 Hz and 600 Hz flicker”
Laycox CA, Thompson R, Haggerty JA, Wilkins AJ, Haigh SM (2024), Perception. PMID 38711325 ↗
- 21
In the same test with an oscillating arm in front of the display, the stroboscopic effect was invisible at dimming frequencies of 1920 or 2160 Hz.
“by presenting an oscillating arm in front of the display”
“When the dimming frequency was 1920 or 2160 Hz, the stroboscopic effect was invisible.”
Guo J, Wei M, Ding R, Zhai D (2025), Journal of the Optical Society of America A. PMID 41411543 ↗
- 22
A 2026 narrative review of LED lighting flicker and seizure risk says the strongest evidence is for visible flicker in photosensitive epilepsy, notably at 3 to 65 Hz with maximal sensitivity at 15 to 25 Hz.
“Critically assess whether light emitting diode (LED) lighting temporal light modulation (TLM) is linked to seizure risk”
“Visible flicker in photosensitive epilepsy, notably in the 3-65 Hz range, with maximal sensitivity at 15-25 Hz, is the strongest evidence.”
Brodaczewski W, Brodaczewski O (2026), Epileptic Disorders. PMID 42635523 ↗
- 23
The same review separates established evidence on visible flicker in photosensitive epilepsy from what it calls weaker evidence on prolonged exposure to invisible flicker, says direct evidence relating typical LED modulation depth to seizures is missing, and concludes that the evidence suggests caution, not causality.
“distinguishing established evidence on visible flicker in photosensitive epilepsy from weaker evidence on prolonged exposure to invisible or supra-critical flicker fusion (CFF) modulation”
“Direct evidence relating typical LED modulation depth to seizures is missing.”
“Evidence suggests caution, not causality.”
Brodaczewski W, Brodaczewski O (2026), Epileptic Disorders. PMID 42635523 ↗
Light and flicker sensitivity in migraine
- 24
In 60 people, 20 without headache and 20 each with light sensitivity between attacks from migraine with or without visual aura, people with migraine did not differ from headache-free controls in how melanopsin and cone signals were combined, but showed an enhanced discomfort response to the combined signal from ipRGCs, the melanopsin-containing light-sensitive cells of the retina.
“the melanopsin-containing intrinsically photosensitive retinal ganglion cells (ipRGCs)”
“We studied 60 people: 20 without headache and 20 each with interictal photophobia from migraine with or without visual aura.”
“people with migraine do not differ from headache-free controls in the manner in which melanopsin and cone signals are combined. Instead, people with migraine demonstrate an enhanced response to integrated ipRGC signals for discomfort.”
McAdams H, Kaiser EA, Igdalova A, et al. (2020), Proceedings of the National Academy of Sciences. PMID 32632006 ↗
- 25
In a model combining the melanopsin and cone luminance action spectra, which best described photophobia thresholds in controls and in people with migraine, melanopsin contributions were about 1.5 times greater than cone luminance.
“A model combining the melanopsin and cone luminance action spectra best described photophobia thresholds in controls and migraineurs; melanopsin contributions were ∼1.5× greater than cone luminance.”
Zele AJ, Dey A, Adhikari P, Feigl B (2021), Cephalalgia. PMID 33040593 ↗
- 26
In 14 people with migraine and 14 healthy controls tested with flicker, those with migraine had significantly elevated discomfort scores, which correlated significantly with the number of years they had had migraine.
“We measured psychophysical thresholds for detection of a flickering target and for the discrimination of suprathreshold flicker contrasts (increment thresholds) in 14 migraineurs and 14 healthy controls”
“Migraineurs also had significantly elevated discomfort scores, and these were significantly correlated with number of years with migraine.”
Karanovic O, Thabet M, Wilson HR, Wilkinson F (2011), Cephalalgia. PMID 21493642 ↗
Evening light and melatonin timing
- 27
The abstract of a 2015 eReader study cites a survey of 1,508 American adults, described then as recent, in which 90% used some type of electronics within an hour before bedtime at least a few nights a week; the survey is not named in the abstract, so cite the original survey rather than this study.
“A representative survey of 1,508 American adults recently revealed that 90% of Americans used some type of electronics at least a few nights per week within 1 h before bedtime.”
Chang AM, Aeschbach D, Duffy JF, Czeisler CA (2015), Proceedings of the National Academy of Sciences. PMID 25535358 ↗
- 28
In that analysis, room light before bedtime shortened melatonin duration by about 90 minutes compared with dim light, and the exposure was room lighting, not a screen.
“Compared with dim light, exposure to room light before bedtime suppressed melatonin”
“shortening melatonin duration by about 90 min”
Gooley JJ, Chamberlain K, Smith KA, et al. (2011), The Journal of Clinical Endocrinology & Metabolism. PMID 21193540 ↗
- 29
With 13 participants viewing tablets set to the highest brightness, melatonin suppression after one hour was not statistically different from zero, but after two hours it was.
“Thirteen participants experienced three experimental conditions”
“tablets-only set to the highest brightness”
“Suppression levels after 1-h exposure to the tablets-only were not statistically different than zero; however, this difference reached significance after 2 h.”
Wood B, Rea MS, Plitnick B, Figueiro MG (2013), Applied Ergonomics. PMID 22850476 ↗
- 30
When the same participants viewed tablets through goggles adding 40 lux of 470 nm light at the cornea, suppression was significantly greater than zero after both one and two hours.
“tablets viewed through clear-lens goggles equipped with blue light-emitting diodes that provided 40 lux of 470-nm light at the cornea”
“Melatonin suppressions after 1-h and 2-h exposures to tablets viewed with the blue light were significantly greater than zero.”
Wood B, Rea MS, Plitnick B, Figueiro MG (2013), Applied Ergonomics. PMID 22850476 ↗
- 31
After 6.5 hours of bright light (about 569 lux), 14 participants each read for two hours, from 21:00 to 23:00, once on a tablet and once in a printed book, and sleep and pre-sleep saliva melatonin did not differ between the two.
“Following a constant bright light exposure over 6.5 hours (~569 lux), 14 participants (six females) read a novel either on a tablet or as physical book for two hours (21:00-23:00).”
“About one week later, experiments were repeated; participants who had read the novel on a tablet in the first experimental session continued reading the same novel in the physical book, and vice versa.”
“There were no differences in sleep parameters and pre-sleep saliva melatonin levels between the tablet reading and physical book reading conditions.”
Rångtell FH, Ekstrand E, Rapp L, et al. (2016), Sleep Medicine. PMID 27539026 ↗
- 32
In 13 healthy boys aged 15 to 17 using LED screens in the evening, blue-blocking glasses significantly attenuated LED-induced melatonin suppression compared with clear lenses, while visually scored sleep stages and behavioral measures collected the next morning were not modified.
“thirteen 15- to 17-year-old healthy male volunteers”
“Compared with clear lenses, BB significantly attenuated LED-induced melatonin suppression in the evening”
“Visually scored sleep stages and behavioral measures collected the morning after were not modified.”
van der Lely S, Frey S, Garbazza C, et al. (2015), Journal of Adolescent Health. PMID 25287985 ↗
How to cite this page
Screen flicker and digital eye strain: what the research shows, and what it does not. trysundown.com, published 28 September 2026, abstracts checked 27 September 2026 and again for the numbers section on 30 September 2026. https://trysundown.com/research/screen-flicker-and-eye-strain
The bibliography and evidence table are archived on Zenodo as DOI 10.5281/zenodo.23007807.
About the publisher
This page is published by Sundown, a Mac menu bar app that warms and dims the display. Its warmth setting runs from 6500 K to 500 K in 100 K steps. None of the cited abstracts mentions the app, and nothing here is a claim about what it does for eyes, sleep or health. This page is a summary of published research, not medical advice.
References
Each abstract was fetched from PubMed and read on 27 September 2026. If a link has rotted or a summary overstates its source, tell us at support@trysundown.com and we will fix it.
- 01
Sheppard AL, Wolffsohn JS (2018). Digital eye strain: prevalence, measurement and amelioration.
BMJ Open Ophthalmology
Review. Symptoms fall into two groups: accommodative or binocular vision stress, and external symptoms linked to dry eye. Critical flicker-fusion frequency, blink rate and completeness are among the objective indices of visual fatigue.
PMID 29963645 ↗ - 02
Portello JK, Rosenfield M, Chu CA (2013). Blink rate, incomplete blinks and computer vision syndrome.
Optometry and Vision Science
21 subjects, 15-minute computer reading task. Symptoms correlated with the share of incomplete blinks. Raising blink rate with an audible tone did not produce a significant change in symptom scores.
PMID 23538437 ↗ - 03
Chu CA, Rosenfield M, Portello JK (2014). Blink patterns: reading from a computer screen versus hard copy.
Optometry and Vision Science
25 subjects, screen versus matched printed text. Blink rates did not differ significantly; incomplete blinks were significantly more frequent on the screen.
PMID 24413278 ↗ - 04
Argilés M, Cardona G, Pérez-Cabré E, Rodríguez M (2015). Blink rate and incomplete blinks in six different controlled hard-copy and electronic reading conditions.
Investigative Ophthalmology & Visual Science
50 participants. Every reading condition lowered blink rate against baseline; only electronic reading raised the share of incomplete blinks.
PMID 26517404 ↗ - 05
Talens-Estarelles C, Cerviño A, García-Lázaro S, Fogelton A, Sheppard A, Wolffsohn JS (2023). The effects of breaks on digital eye strain, dry eye and binocular vision: testing the 20-20-20 rule.
Contact Lens and Anterior Eye
29 symptomatic computer users. Break reminders reduced eye strain and dry eye symptoms; the effect was not maintained a week after stopping, and no change was observed in any ocular surface or tear film parameter.
PMID 35963776 ↗ - 06
Singh S, Keller PR, Busija L, et al. (Cochrane Review) (2023). Blue-light filtering spectacle lenses for visual performance, sleep, and macular health in adults.
Cochrane Database of Systematic Reviews
17 randomised trials. Blue-light filtering lenses may not attenuate eye strain with computer use over short follow-up; no clinically meaningful difference in critical flicker-fusion frequency; sleep effects indeterminate; no trial measured serum melatonin.
PMID 37593770 ↗ - 07
Rosenfield M, Li RT, Kirsch NT (2020). A double-blind test of blue-blocking filters on symptoms of digital eye strain.
Work
24 subjects, 20-minute tablet reading. Symptoms rose after the task, with no significant difference between blue-blocking and control lenses.
PMID 32007978 ↗ - 08
Vera J, Redondo B, Ortega-Sanchez A, et al. (2023). Blue-blocking filters do not alleviate signs and symptoms of digital eye strain.
Clinical and Experimental Optometry
23 young adults, 30-minute computer reading. A blue-blocking filter did not significantly alter orbicularis oculi muscle activity or visual symptoms.
PMID 35057697 ↗ - 09
Yoshimoto S, Garcia J, Jiang F, Wilkins AJ, Takeuchi T, Webster MA (2017). Visual discomfort and flicker.
Vision Research
Discomfort from flicker rose with deviations from natural temporal amplitude spectra and was strongly affected by the phase spectrum.
PMID 28709920 ↗ - 10
Yoshimoto S, Jiang F, Takeuchi T, Wilkins AJ, Webster MA (2020). Visual discomfort from flicker: effects of mean light level and contrast.
Vision Research
The relative comfort of one flicker waveform held only at photopic light levels; discomfort tracked how natural the flicker looked across contrasts and light levels.
PMID 32474213 ↗ - 11
Patterson Gentile C, Aguirre GK (2020). A neural correlate of visual discomfort from flicker.
Journal of Vision
10 headache-free people and 10 with migraine with aura. The amplitude of the visual cortical response to flicker related linearly to reported discomfort; no substantive group differences.
PMID 32667963 ↗ - 12
Jiang Y, Zhou K, He S (2007). Human visual cortex responds to invisible chromatic flicker.
Nature Neuroscience
fMRI showed many visual cortical areas distinguish chromatic flicker at 25 Hz or higher, perceived as one fused colour, from a non-flickering control.
PMID 17396122 ↗ - 13
Laycox CA, Thompson R, Haggerty JA, Wilkins AJ, Haigh SM (2024). Flicker and reading speed: effects on individuals with visual sensitivity.
Perception
Reading under LED light was slowest with 60 Hz and 600 Hz flicker. The effect of flicker on reading speed was larger in the low pattern-glare group.
PMID 38711325 ↗ - 14
Stanikunas R, Bliumas R, Jocbalyte K, Novickovas A, Soliunas A (2025). Color perception of flickering light in the phantom array.
Journal of the Optical Society of America A
During a saccade a flickering light is seen as a sequence of dashes; perceived colour sequences varied between subjects, and no distortion appeared without eye movement.
PMID 40793556 ↗ - 15
Brodaczewski W, Brodaczewski O (2026). Temporal light modulation from LED lighting and seizure risk: a focused review of visible flicker, invisible modulation, and evidence gaps.
Epileptic Disorders
Narrative review; first author affiliation listed as Grena Limited, London, UK. Strongest evidence concerns visible flicker in photosensitive epilepsy; direct evidence linking typical LED modulation depth to seizures is missing. Calls for double-masked studies of headache, eyestrain and discomfort against modulation depth.
PMID 42635523 ↗ - 16
Guo J, Wei M, Ding R, Zhai D (2025). Characterization of the stroboscopic effect for OLED displays using the PWM dimming method.
Journal of the Optical Society of America A
An oscillating arm was presented in front of an OLED smartphone display. The stroboscopic effect was visible at 480 and 960 Hz dimming below 40 cd/m², and invisible at 1920 or 2160 Hz.
PMID 41411543 ↗ - 17
McAdams H, Kaiser EA, Igdalova A, et al. (2020). Selective amplification of ipRGC signals accounts for interictal photophobia in migraine.
Proceedings of the National Academy of Sciences
60 people. Migraine did not change how melanopsin and cone signals combine, but amplified the discomfort response to the integrated ipRGC signal.
PMID 32632006 ↗ - 18
Zele AJ, Dey A, Adhikari P, Feigl B (2021). Melanopsin hypersensitivity dominates interictal photophobia in migraine.
Cephalalgia
Melanopsin contributions to photophobia thresholds were about 1.5 times those of cone luminance; migraineurs between attacks had lower photophobia thresholds.
PMID 33040593 ↗ - 19
Karanovic O, Thabet M, Wilson HR, Wilkinson F (2011). Detection and discrimination of flicker contrast in migraine.
Cephalalgia
14 migraineurs and 14 controls. Migraineurs had significantly elevated discomfort scores, significantly correlated with years with migraine.
PMID 21493642 ↗ - 20
Wilkins AJ, Haigh SM, Mahroo OA, Plant GT (2021). Photophobia in migraine: a symptom cluster?.
Cephalalgia
Review arguing that a cortical mechanism explains discomfort from light, flicker, patterns and colour in migraine more parsimoniously than ipRGC activity alone.
PMID 33990148 ↗ - 21
Chang AM, Aeschbach D, Duffy JF, Czeisler CA (2015). Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness.
Proceedings of the National Academy of Sciences
Compared with a printed book, reading a light-emitting eBook before bed lengthened time to fall asleep, reduced melatonin secretion, delayed the circadian clock and reduced next-morning alertness.
PMID 25535358 ↗ - 22
Gooley JJ, Chamberlain K, Smith KA, et al. (2011). Exposure to room light before bedtime suppresses melatonin onset and shortens melatonin duration in humans.
The Journal of Clinical Endocrinology & Metabolism
116 healthy adults aged 18 to 30. Room light (under 200 lux) before bedtime, against dim light, suppressed melatonin, delayed its onset in nearly all individuals and shortened its duration.
PMID 21193540 ↗ - 23
Wood B, Rea MS, Plitnick B, Figueiro MG (2013). Light level and duration of exposure determine the impact of self-luminous tablets on melatonin suppression.
Applied Ergonomics
13 participants, tablets at highest brightness. Suppression after one hour was not statistically different from zero; after two hours it was.
PMID 22850476 ↗ - 24
Schöllhorn I, Stefani O, Lucas RJ, Spitschan M, Slawik HC, Cajochen C (2023). Melanopic irradiance defines the impact of evening display light on sleep latency, melatonin and alertness.
Communications Biology
72 healthy males. With luminance and colour held constant, sleep latency, melatonin concentration and melatonin onset changed dose-dependently with melanopic irradiance; subjective alertness did not.
PMID 36854795 ↗ - 25
Rångtell FH, Ekstrand E, Rapp L, et al. (2016). Two hours of evening reading on a self-luminous tablet vs. reading a physical book does not alter sleep after daytime bright light exposure.
Sleep Medicine
14 participants after 6.5 hours of bright daytime light. No differences in sleep or pre-sleep saliva melatonin between tablet and book. Authors call for larger studies.
PMID 27539026 ↗ - 26
van der Lely S, Frey S, Garbazza C, et al. (2015). Blue blocker glasses as a countermeasure for alerting effects of evening light-emitting diode screen exposure in male teenagers.
Journal of Adolescent Health
13 boys aged 15 to 17. Blue-blocking glasses attenuated LED-screen melatonin suppression against clear lenses; sleep stages and next-morning measures were not modified.
PMID 25287985 ↗ - 27
Lin CW, Yeh FM, Wu BW, Yang CH (2019). The effects of reflected glare and visual field lighting on computer vision syndrome.
Clinical and Experimental Optometry
Two-hour visual task. Fatigue scores were higher with glare and lower with a glare-free display; the rise in scores was smaller with uniform supplementary lighting. Critical fusion frequency was the only visual function parameter that decreased significantly.
PMID 30805993 ↗ - 28
Buchner A, Baumgartner N (2007). Text-background polarity affects performance irrespective of ambient illumination and colour contrast.
Ergonomics
Proofreading was better with dark text on a light background, in darkness and office light alike; self-reported eyestrain and headache did not vary between conditions.
PMID 17510822 ↗ - 29
Rodán A, Fernández-López A, Vera J, Montoro PR, Redondo B, Prieto A (2025). Too bright to focus? Influence of brightness illusions and ambient light levels on the dynamics of ocular accommodation.
Vision (Basel)
32 young adults. Visual comfort decreased for stimuli made to look brighter by illusion, especially under low ambient light.
PMID 41133605 ↗