Most evening advice is about what you do at 10pm. Almost all of the effect was set by what you drank at 2pm.
The short answer
Move two things earlier. Keep one thing constant.
Across 24 studies caffeine cut total sleep by 45 minutes; 400 mg six hours before bed cost over an hour the drinkers never noticed. Alcohol buys the first half of the night and sells the second. Two standard rules don’t hold, though: evening exercise didn’t harm sleep across 23 studies, and a four-hour-later dinner changed nothing.
The four rules
| Habit | The rule | The evidence |
|---|---|---|
| Caffeine | Last cup ~9 hours out | 24-study meta-analysis: 8.8 hours for a 107 mg cup, 13.2 for a 217.5 mg pre-workout |
| Alcohol | Less, and earlier | Delays the first REM period at every dose; more waking later on |
| Evening exercise | Finish an hour out | 23 studies: no harm, except vigorous work ending ≤1 hour before bed |
| Bed and wake times | Same clock, weekends too | In 60,977 adults the four most regular fifths had 20–48% lower all-cause mortality |
Caffeine is later than you think
Twelve normal sleepers took 400 mg of caffeine — two to three cups — at bedtime, three hours before, or six hours before. Even the six-hour dose cut total sleep by over an hour. The 2023 pooled analysis: 45 minutes less sleep, efficiency down 7%, nine more minutes to fall asleep, twelve more awake. Full caffeine timing here.
You can’t catch this by feel: the six-hour group lost over an hour of sleep and reported nothing unusual. The cost lands at 3am, never billed to the cup.
Alcohol: the trade you’re making
A nightcap works on the part you notice — faster to sleep, more deep sleep early. Then the bill: the first REM period is delayed at every dose, total-night REM drops at moderate and high doses, the back half breaks up.
Two rules you can drop
Evening exercise. Pooled across 23 studies it slightly improved sleep: deep sleep up 1.3 percentage points, REM latency up 7.7 minutes, stage 1 down 0.9. Harm showed only with vigorous work ending within an hour of bed.
The 9pm dinner. Twenty adults ate at 6pm and at 10pm in a crossover sleep study. Sleep time, efficiency, latency and stage distribution came out identical. The late meal did move overnight glucose and fat oxidation — the case against it is metabolic, not sleep.
Who should skip these
- On prescription medication. Valerian, passionflower and hemp interact with sedatives, blood thinners and some antidepressants. Ask a provider first.
- Pregnant or nursing. Skip herbal sedatives and hemp; ask about caffeine limits.
- If it has lasted weeks, that’s a provider, not a tea.
Before you buy anything
Do the free things first: shift the last coffee, drop the nightcap, hold Saturday’s wake time. Better evidenced than anything on the shelf. Still curious in a month? Start with what CBD and melatonin each do.
Questions people ask
I can’t stop caffeine nine hours out. What’s next best?
Cut the dose, not just the hour. The 8.8-hour figure is for a 107 mg cup; a bigger cup needs a bigger gap. No trial has tested that swap.
Does a nightcap ever help?
It gets you to sleep and costs you the back half of the night. That’s the trade, every time.
Is herbal tea doing anything?
The warmth, the ritual, and the fact it isn’t a beer. Chamomile trials are small and mixed.
Weekends too?
That’s where regularity is won or lost. A steady wake time beats a steady bedtime.
Sources
- Caffeine Effects on Sleep Taken 0, 3, or 6 Hours before Going to Bed — J Clin Sleep Med, 2013.
- The effect of caffeine on subsequent sleep — Sleep Medicine Reviews, 2023;69:101764.
- Alcohol and Sleep I: Effects on Normal Sleep — Alcohol Clin Exp Res, 2013.
- Effects of Evening Exercise on Sleep in Healthy Participants — Sports Medicine, 2019;49:269–287.
- Effects of Dinner Timing on Sleep Stage Distribution — Nature and Science of Sleep, 2021;13:601–612.
- Sleep regularity is a stronger predictor of mortality risk than sleep duration — Sleep, 2024;47(1):zsad253.
How we make money. Some links on this page are affiliate links. If you buy through one, we may earn a commission at no extra cost to you. It does not change what we write — see our affiliate disclosure.
These statements have not been evaluated by the Food and Drug Administration. CBD products are not intended to diagnose, treat, cure or prevent any disease. Serene Leaf does not physically test products; our assessments are based on published research, third-party lab reports and publicly available product information.
This article is information, not medical advice. Talk to a qualified healthcare provider before starting any supplement — especially if you are pregnant, nursing, taking medication or managing a diagnosed condition. 21+ only.