What the Research Says
What the Sleep Research Actually Shows
Hundreds of sleep studies agree on a short list of high-leverage interventions. Here's what the evidence actually supports — and what it doesn't.
The Problem with Sleep Advice
Most sleep advice is one of two things: obvious (go to bed at the same time every night) or unfounded (take this supplement, follow this routine). The research literature is more useful and more specific than either.
After reviewing the peer-reviewed evidence, a few findings stand out for their consistency across study designs, populations, and effect sizes.
What the Evidence Consistently Supports
Consistent sleep and wake timing is the strongest single intervention. Across dozens of controlled studies, irregular sleep timing — not just short sleep — predicts worse mood, cognition, and metabolic markers. The circadian rhythm is not forgiving. One night of a shifted schedule measurably degrades next-day performance even when total sleep hours are held constant.
Light exposure drives circadian alignment. Morning bright light (1000+ lux, ideally outdoors) advances the circadian phase and makes falling asleep at your target time easier. Evening blue light does the reverse. The effect sizes here are large — 30 minutes of morning outdoor light shows measurable effects on sleep onset in controlled trials.
Core body temperature predicts sleep onset. Sleep begins as core temperature drops. A warm bath or shower 1–2 hours before bed accelerates this drop via peripheral vasodilation, consistently reducing sleep onset latency in RCTs. Cold bedroom temperatures (16–19°C) have a similar effect.
Caffeine's half-life is longer than most people assume. It's approximately 5–6 hours in most adults, but with significant individual variation based on CYP1A2 genotype. A 200mg dose at 2pm leaves roughly 100mg active at 8pm. Studies measuring sleep architecture (not just subjective quality) show that late caffeine reduces slow-wave sleep even when subjects report sleeping fine.
What the Evidence Is Weaker On
Sleep tracking accuracy varies significantly by device. Wrist-based actigraphy tends to overestimate sleep and underestimate wake, especially in poor sleepers. Subjective sleep quality ratings often diverge from objective measures — which makes self-experimentation particularly valuable here.
Supplements show mixed results. Melatonin has strong evidence for circadian phase-shifting (jet lag, shift work) but weaker evidence as a general sleep quality enhancer in people without circadian disruption. Magnesium glycinate shows some signal but the effect sizes in quality RCTs are modest.
"Sleep hygiene" bundles are hard to study cleanly. Most RCTs test multiple interventions together, making it impossible to isolate which component drives the effect.
Why Individual Variation Matters
The population-level averages mask enormous individual differences. Caffeine sensitivity, chronotype, optimal sleep duration, and response to light all vary meaningfully between people. The literature gives you a good prior — it's worth testing whether the high-leverage interventions actually move your personal sleep metrics.