Evening screens shift bedtimes. The blue light is the smaller part of it.

Blue light before bed has a real, measured biological effect: in a controlled comparison, people reading a light-emitting e-reader in the hours before bedtime took longer to fall asleep, secreted less melatonin, showed a later circadian phase and were less alert the next morning than when they read a printed book [s1]. But when researchers follow people through ordinary evenings and ask what actually changes their sleep from one night to the next, the answer that comes back is bedtime — not biology.

Two studies published in 2026 make that case from different directions, and neither of them contradicts the laboratory work. They measure a different thing.

The laboratory result, stated precisely

The e-reader experiment compared reading on a light-emitting device with reading a printed book in the hours before bed [s1]. The device condition produced a longer time to fall asleep, reduced evening sleepiness, suppressed melatonin secretion, a phase delay in the circadian clock, and reduced next-morning alertness [s1]. The mechanism is not in dispute: short-wavelength-enriched light has been shown experimentally to produce alerting effects, suppress melatonin and phase-shift the biological clock [s1].

Context for how universal the exposure is: the paper cites a representative survey of 1,508 American adults in which 90% reported using some type of electronics at least a few nights per week within one hour before bedtime [s1].

That is a clean experimental demonstration that evening screen light can do something to human circadian physiology. What it does not establish is how much of the real-world association between screens and worse sleep is attributable to that mechanism, as opposed to the simpler fact that people who are on a device at 11pm are not asleep at 11pm.

The within-person test

A systematic review and meta-analysis in JAMA Pediatrics pooled the studies that ask the more useful question: on days when a given person used screens more than usual, did that same person sleep worse than usual? Twenty-five studies covering 4,562 participants aged 3 to 25 were included, from a literature search running to 22 August 2025 [s2].

The pooled within-person correlation between screen time and sleep onset was small, positive and statistically significant — r = 0.079 (95% CI, 0.010–0.149; P = .03) — meaning later bedtimes on days with more screen use [s2]. On every other outcome the analysis found nothing: no significant within-person association with total sleep time, sleep onset latency, sleep efficiency, wake after sleep onset, or subjective sleep quality [s2].

The one moderator that mattered was timing. The association between screen time and subjective sleep quality differed significantly by when the screen use happened: screen time after bedtime showed a stronger negative correlation (r = −0.092) than daily (r = −0.026) or evening (r = −0.005) assessments (P = .007) [s2]. Screen type made no significant difference, and neither did the method used to assess sleep [s2].

Read that pattern carefully. If light exposure were the dominant pathway, evening screen use — the window in which light suppresses melatonin — should be where the effect lives. Instead the effect concentrated in screen use after the person had already gone to bed, which is the window in which screens displace sleep rather than merely precede it.

The arousal hypothesis, tested directly

The other common explanation is arousal: that games, social feeds and messages wind people up physiologically. A New Zealand study set out to measure it, recording screen use with wearable cameras over four nights in 70 young people aged 11 to 14.9 years, 37% of them Māori and 42% female, with heart rate captured by a Fitbit Inspire 2 [s3].

The result ran the opposite way from the hypothesis. Median heart rate was lower during screen use than during non-screen activities: 83 beats per minute (IQR 77–91) against 93 (IQR 87–100) [s3]. Social media use closest to bedtime was associated with a slightly lower heart rate than other screen activities (−3 BPM, 95% CI −5 to −1), while communication activities were associated with a slightly higher one (3 BPM, 95% CI 1 to 5) [s3].

Heart rate in the two hours before bed was not associated with sleep outcomes, with one exception: sleep latency increased by 9 minutes (95% CI 3 to 14) for every 10 BPM increase in heart rate [s3]. The authors' conclusion is that evening screen use may function as a relaxation tool for young people, and that the relationship between screen use and sleep is more likely driven by changes in sleep timing than by physiological arousal from screen exposure [s3].

The limits worth stating

The within-person meta-analysis rests on 25 studies and 4,562 participants, which is not a large literature for a question this heavily discussed, and its correlations are small enough that publication and measurement noise could move them [s2]. The heart-rate study is 70 adolescents over four nights in one country, with heart rate from a consumer wearable rather than from electrocardiography [s3]. And the e-reader experiment was a controlled comparison of two reading conditions, which is exactly what makes its mechanism finding clean and exactly what limits its generalisation to a person scrolling in bed [s1].

None of the three studies measured what happens over years, and none of them tested an intervention in which screens were removed and sleep improved.

Where the evidence points

The mechanism is real and the field effect is mostly about the clock on the wall. Light from a device can suppress melatonin and delay circadian timing under controlled conditions [s1]. In day-to-day life, more screen use predicts a later sleep onset and very little else, and the clearest harm signal attaches to screen use after bedtime rather than to screen use in the evening [s2] [s3].

That distinction changes what the evidence can and cannot support. It supports the claim that device use late at night is associated with going to sleep later. It does not, on this evidence, support the stronger claim that filtering the light is what would fix it.

Sources

• Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness — Proceedings of the National Academy of Sciences, 2014-12-22
• Within-Person Association Between Daily Screen Use and Sleep in Youth: A Systematic Review and Meta-Analysis — JAMA Pediatrics, 2026-03-02
• Screens, Teens, and Sleep: Is the Impact of Nighttime Screen Use on Sleep Driven by Physiological Arousal? — Journal of Sleep Research, 2026-01-20