Does sleeping with lights on or living with night-time light harm your health?

Your brain's clock reads light through the eyes and uses it to time sleep, hormones and metabolism. Light at the wrong time sends the wrong message.

How little it takes. In 55 young adults, evening light of 10, 30 and 50 lux delayed the melatonin rise by 22, 77 and 109 minutes. Half the maximum effect came at about 25 lux, and people differed more than tenfold in sensitivity⁶. In 116 volunteers, ordinary room light before bed delayed melatonin onset in 99% and shortened melatonin duration by about 90 minutes compared with dim light⁷. For screens specifically, see Evening Screen Exposure.

Light during sleep. In a randomised trial, 20 healthy adults spent a night in dim light or in 100 lux of room light. Room light raised heart rate, lowered heart rate variability and raised next-morning insulin resistance¹. In a small lab study, ten sleepers slept more lightly, with more arousals, when the lights were on⁸.

In real bedrooms. Among 528 older Japanese adults, bedrooms averaging 3 lux or more went with higher body weight, waist size and triglycerides, and with obesity (odds ratio 1.89)². Among 863 adults free of depression, those with bedrooms of 5 lux or more had 1.89 times the risk of developing depressive symptoms over about two years³. In 43,722 US women, sleeping with a light or television on went with a 33% higher risk of becoming obese⁹. In 84,790 people wearing light sensors, the brightest tenth of nights went with a 53% higher risk of type 2 diabetes⁴.

Streetlight and shift work. A meta-analysis of 7 cross-sectional studies (577,932 people) linked light at night to 22% greater odds of sleep problems, with indoor light mattering more than outdoor light¹⁰. IARC classes night shift work as "probably carcinogenic" (Group 2A)¹¹.

Practical targets. An expert consensus recommends light no brighter than 10 'melanopic' lux (a measure of how strongly light drives the body clock) for at least 3 hours before bed, and a sleep room as dark as possible, under 1 melanopic lux⁵. In a simulated hospital night, eye masks with earplugs improved sleep and raised melatonin¹². For other habits, see Sleep Hygiene.

Randomised lab trial. Mason and colleagues randomised 20 healthy adults to a second night in dim light (under 3 lux, n = 10) or a second night under overhead room light at 100 lux (n = 10), after a dim first night. After the room-light night, HOMA-IR, a measure of insulin resistance, rose about 15% against a 4% fall in the dim group (P = 0.018). Heart rate was higher, heart rate variability lower, and slow-wave and REM sleep took up a smaller share of the night. A higher sympathetic share of heart rhythm correlated with a larger early insulin response (R = 0.65 across both groups). Glucose stayed normal because insulin rose to compensate. It was a single night in healthy, normal-weight volunteers, so lasting effects are unknown¹.

Sleep depth. In Cho and colleagues' polysomnography study, ten healthy sleepers slept once with the lights off and once with them on. Lights-on sleep had more light stage 1 sleep, less slow-wave sleep and a higher arousal index, with lower slow-oscillation, delta and spindle power⁸.

Evening and sleep-hours light. Gooley and colleagues analysed 116 volunteers aged 18 to 30 living in a research centre. Compared with dim light (under 3 lux), room light (under 200 lux) in the 8 hours before bed delayed melatonin onset in 99.0% of people and shortened melatonin duration by about 90 minutes. Room light during the usual hours of sleep suppressed melatonin by more than 50% in 85% of trials⁷. Phillips and colleagues exposed 55 people to 10 to 2,000 lux for 5 evening hours. The light level giving half the maximal melatonin suppression was 24.6 lux for the group, but ranged from 6 lux in the most sensitive person to 350 lux in the least⁶.

Cohorts. In the HEIJO-KYO study, bedroom light was measured every minute for two nights. Of 528 adults (mean age 72.8), the 145 sleeping at an average of 3 lux or more had higher BMI (23.3 vs 22.7), waist (84.9 vs 82.8 cm) and triglycerides (119.7 vs 99.5 mg/dl), and lower HDL cholesterol, than the 383 in dimmer rooms. Odds ratios were 1.89 for obesity by BMI and 1.72 for dyslipidemia, independent of urinary melatonin². A follow-up of 863 adults without depressive symptoms (median 24 months) found that 73 developed them. Those in bedrooms averaging 5 lux or more (153 people) had a hazard ratio of 1.89 (1.13 to 3.14), or 1.72 (1.03 to 2.89) after adjusting for hypertension, diabetes and sleep³. In the Sister Study, 43,722 women free of cancer and cardiovascular disease were followed for 5.7 years on average. Sleeping with a television or light on in the room went with a relative risk of 1.33 (1.13 to 1.57) for incident obesity and 1.17 (1.08 to 1.27) for gaining 5 kg or more, compared with no light. Light exposure was self-reported⁹. In UK Biobank, 84,790 people wore light sensors for a week and were followed for about 8 years, with 1,997 new cases of type 2 diabetes. Compared with the darkest half of nights, hazard ratios were 1.29 (1.14 to 1.46), 1.39 (1.24 to 1.57) and 1.53 (1.32 to 1.77) across progressively brighter night-light groups, independent of genetic risk⁴.

Meta-analysis, outdoor light and shift work. A meta-analysis of 7 cross-sectional studies (577,932 people) found higher light at night went with 22% greater odds of sleep problems (summary odds ratio 1.22, 1.13 to 1.33), more strongly for indoor light (1.74) than outdoor light (1.19)¹⁰. IARC classified night shift work as "probably carcinogenic" on limited human evidence for breast, prostate, colon and rectal cancer, sufficient animal evidence and strong mechanistic evidence in experimental animals¹¹.

Null and conflicting findings. Not every result lines up. Urinary melatonin did not differ between the bright-bedroom and dim-bedroom groups, so melatonin suppression may not explain the metabolic links². In the lab trial a single night of light changed insulin handling but not glucose¹. The outdoor-light association is smaller than the indoor one, and every study in that meta-analysis was cross-sectional¹⁰. We found no randomised trial showing that darkening bedrooms improves long-term health.

Practical guidance and its limits. The recommendations of Brown and colleagues are an expert consensus, not trial results: in daytime at least 250 melanopic lux at the eye, from 3 hours before bed no more than 10, and in the sleep room no more than 1 (10 for night tasks). Melanopic lux counts how strongly light drives the clock cells in the eye, and it differs from the reading on a phone light meter⁵. In 14 healthy volunteers in a simulated intensive care unit, earplugs and eye masks gave more REM sleep, less arousal and higher melatonin¹².

Why caution is warranted. The cohort links are modest and observational. People who sleep with the TV on, sleep badly or are depressed may differ in other ways, and reverse causation is possible, because low mood or insomnia may lead people to leave lights on. The lab trial was small and short.

Bottom line (confidence 3). Bright bedrooms and bright nights likely impair sleep and metabolic health and are linked to mood, with a plausible mechanism and a consistent direction across designs. The size of the real-world effects is uncertain. Darkening the bedroom is cheap and low-risk.