The changes with the strongest research behind them are surprisingly few: keep a consistent wake time, get bright light early in the day, stop caffeine at least six hours before bed, keep the bedroom cool and dark, and — if you have been sleeping badly for months — use cognitive behavioral therapy for insomnia rather than a longer habit checklist. Almost everything else in the standard advice list is either a small effect, a preference, or a rule that only matters if you are breaking it badly.
That short list matters because most “how to sleep better” advice is written for someone whose nights are fully within their own control. If you are up twice with a baby, or waking at 3am with your heart going and the sheets damp, a fifteen-step wind-down routine is not the missing piece. What follows is the evidence, ranked, and then what to do with it when the interruptions are not optional.
What actually improves sleep, in order of evidence?
1. A consistent wake time. This is the highest-leverage change most people never make. Windred and colleagues, publishing in SLEEP in 2024, analyzed accelerometer data from more than 60,000 UK Biobank participants and found that the regularity of the sleep-wake pattern predicted mortality risk more strongly than total sleep duration did. Regularity is anchored by when you get up, not when you go to bed — bedtime follows sleepiness, but wake time is a decision. We go deeper on that trade-off in sleep consistency vs. duration.
2. Bright light early, dim light late. Your circadian clock is set primarily by light. Wright and colleagues (Current Biology, 2013) took participants camping for a week with no electric light and found their internal melatonin timing shifted almost two hours earlier, aligning to natural dawn. Chang and colleagues (PNAS, 2015) showed the reverse: reading on a light-emitting device before bed suppressed melatonin, delayed sleep onset, and left readers less alert the next morning. Practically: get outside within an hour of waking (morning sunlight does more than any supplement), and dim things down in the last hour.
3. Caffeine timing. Drake and colleagues (Journal of Clinical Sleep Medicine, 2013) gave participants 400mg of caffeine at 0, 3, or 6 hours before bedtime. Even the six-hour-before dose reduced total sleep time by more than an hour compared to placebo — and participants did not reliably notice. The dose you cannot feel is still the dose that is costing you sleep. More on where the line sits in caffeine and sleep.
4. A cool, dark bedroom. Okamoto-Mizuno and Mizuno (Journal of Physiological Anthropology, 2012) reviewed the thermal environment literature and found that heat exposure increases wakefulness and suppresses slow-wave and REM sleep, with humidity compounding the effect. Core body temperature has to fall for sleep to begin, and a warm room blocks that. Haghayegh and colleagues (Sleep Medicine Reviews, 2019) found the related trick works too: a warm bath or shower one to two hours before bed shortened sleep onset by about ten minutes on average, because the peripheral warming accelerates heat loss afterward.
5. CBT-I, if this has been going on for months. Trauer and colleagues (Annals of Internal Medicine, 2015) pooled 20 randomized trials of cognitive behavioral therapy for chronic insomnia and found it reduced time to fall asleep by about 19 minutes and time awake after falling asleep by about 26 minutes, with effects that held at follow-up. That is the first-line treatment in the clinical guidelines — ahead of medication. See how to treat insomnia without medication for what a course actually involves.
What is overrated?
Not wrong, exactly — just smaller than its reputation.
Irish and colleagues (Sleep Medicine Reviews, 2015) went through the standard sleep hygiene recommendations one at a time and found the evidence base uneven. Some items rest on studies of extremes that do not map onto normal life. Their conclusion was that sleep hygiene functions as a reasonable foundation, not as a treatment. We unpacked which pieces hold up in what is sleep hygiene.
Alcohol deserves a specific mention because it is so often used as a sleep aid. Ebrahim and colleagues (Alcoholism: Clinical and Experimental Research, 2013) reviewed the effects on normal sleep: alcohol does shorten sleep onset and increases slow-wave sleep in the first half of the night, then produces marked disruption and REM suppression in the second half. It is a sedative that trades the back half of your night for the front half. Details in alcohol and sleep.
Exercise is genuinely helpful but often oversold as a fix. Kredlow and colleagues (Journal of Behavioral Medicine, 2015) meta-analyzed the literature and found regular exercise produced small-to-moderate improvements in sleep quality and onset latency — real, worth having, and not the same category of effect as fixing your wake time. More in exercise and sleep.
And tracking, honestly, sits here too. Watching a number every morning does not by itself change anything, and for some people it makes things worse — the reason we do not show a 0-100 score anywhere in Mendtide, and why we wrote about why sleep scores cause anxiety.
How do you sleep better with a newborn or young child?
You cannot prevent the wake-ups. So the goal moves: protect the quality of the sleep around them.
Hunter, Rychnovsky and Yount (JOGNN, 2009) reviewed maternal sleep in the postpartum period and found the defining feature is not shortened total sleep so much as severe fragmentation — sleep is broken into pieces too short to cycle properly through deep and REM stages. That distinction changes what helps. Chasing a longer night is mostly out of your hands. Reducing the cost of each interruption is not.
That means keeping the return-to-bed as low-stimulation as you can: dim light rather than overhead light, no phone, no clock-watching. It means holding your wake time roughly steady even after a bad night, because a wildly variable schedule adds a second problem on top of the first. And it means treating daytime light exposure as non-negotiable — it is the one circadian input a broken night cannot take from you.
If you are lying awake after the baby has gone back down, that is a different problem with its own literature — see postpartum insomnia. And if you are trying to work out whether the accumulated deficit is something you can repay, sleep debt for parents covers what recovery realistically looks like.
How do you sleep better through perimenopause?
Baker and colleagues (Nature and Science of Sleep, 2018) reviewed sleep across the menopausal transition and found insomnia complaints rise substantially — driven by vasomotor symptoms, shifting progesterone and estrogen, and increased rates of mood disturbance and sleep-disordered breathing during this window.
The practical implication is that the temperature lever, which is a minor tweak for most people, becomes a primary one here. Cooler room, layered bedding you can shed without fully waking, breathable fabrics. The same 2012 thermal review applies with more force when your own thermoregulation is being disrupted from the inside — see best bedroom temperature for sleep.
The other implication is the 3am wake-up, which is close to universal in this group and is usually a collision of a cortisol rise, a light sleep stage, and a vasomotor event. It is not a character flaw and it is not usually insomnia in the clinical sense. Why you wake at 3am explains the mechanism, and perimenopause and sleep covers the hormonal picture in full.
What should you actually change first?
One thing. Pick the highest item on the evidence list that you are currently not doing, and hold it for two weeks before adding anything else.
For most people that is the wake time. For people who are already regular, it is usually light — getting outside early, dimming late. For people who have been sleeping badly for three months or more, it is starting CBT-I rather than adding another habit. Stacking five changes at once means you learn nothing about which one worked, and you give up when the composite is too hard to maintain.
If your sleep is disrupted by pain, breathing, or a symptom pattern that has not shifted with any of this, that is a conversation for a clinician rather than a habit change.
The calm version
There are only a handful of changes with real evidence behind them, and you do not need all of them. A steady wake time, light early, caffeine earlier than feels necessary, a cool dark room — that is most of the available effect. If your nights are being interrupted by a baby or a hormonal shift, the goal is not an unbroken night; it is protecting what happens around the breaks. Mendtide reads your Apple Health sleep data and points at the one thing worth changing next, with the research attached, rather than handing you a score to worry about.
You do not need a perfect night. You need a slightly better next two weeks, and one change small enough that you will actually still be doing it by then.