If a fixed nighttime boundary fits your routine, you can download 9to9 from the App Store and choose the apps and websites you want quiet from 9 PM to 9 AM.
The question gets asked like there is a number — thirty minutes, an hour, two hours — and the honest answer is that the research does not give you one, and anyone who tells you a precise figure is rounding off their uncertainty. What the evidence does support is a direction: the closer the screen use is to the moment of sleep, the more it is associated with worse outcomes — later bedtimes, shorter sleep, worse reported quality. So the right question is not “exactly how long before bed” but “what should the last stretch of the night be for,” and the answer to that is clearer than the number.
The question you are probably actually asking
Worth naming what the question usually means in practice, because it is rarely abstract. Most people asking “how long before bed” are not optimizing a sleep protocol — they are noticing a pattern: the phone is the last thing at night and the first thing in the morning, and the nights keep running later than intended. If that is the real question, the answer is less about minutes than about the feed's role: the scroll should not be the thing the evening ends in, and the simplest way to ensure that is a boundary that ends it for you. The “how long” is the mechanism; the “what ends the night” is the point.
Why there is no universal number
The reason no specific cutoff has been established is that the harm is not a threshold — it is not that 45 minutes is safe and 44 is not. The association runs through several different mechanisms, each with its own relationship to time. The displacement effect — the scroll eating the hour — matters most in the last stretch before sleep, when the hour is the one you cannot afford. The arousal effect — the content keeping you alert — matters in proportion to how stimulating the content is, which varies wildly. And the light effect depends on brightness and duration in ways that a single cutoff cannot capture.
An expert consensus statement on screen use and sleep is careful about exactly this: the effects depend on content, timing, light, age, and context — which is the scientific way of saying there is no one number because the variables differ too much from person to person. What is consistent is the direction: screen use in the period close to sleep is the version most associated with the worse outcomes.
Sources: The impact of screen use on sleep health across the lifespan: consensus statement (PMC)
The useful way to think about it: not a number, a zone
A more honest and more usable frame than a cutoff is a zone: the last stretch of the night — roughly the final hour before you intend to sleep — is the period where the scroll does its damage, and the period that most needs to belong to something else. Within that zone, the question is not “can I use the phone” but “can the phone be the last thing,” and the answer to that is no. The scroll in the final stretch is the specific version of the habit the research most consistently links to the bad outcomes; the phone earlier in the evening is a much smaller problem.
This reframing matters because it changes what “on time” means. Putting the phone down 30 minutes before a 1 AM bedtime is not the win the question implies — the win is the 1 AM bedtime not happening because the feed did not hold you past the intended one. The boundary protects the bedtime; the bedtime is the point.
What the research does suggest about timing
If there is no number, there is at least a shape. The studies that find the strongest associations tend to be the ones measuring use in bed or in the immediate pre-sleep period — the scroll that ends at the moment of sleep, or continues into it. Use earlier in the evening is a much weaker signal. That pattern — strongest at the pillow, weaker earlier — is itself a kind of answer: the problem is not the phone at night in general, it is the phone as the thing the night ends in. The practical reading is that the boundary does not need to be hours before bed; it needs to be before the feed becomes the last thing.
A workable default
If you want a number to start from — and a starting number is worth having even if it is not universal — aim for 30 to 60 minutes before the time you intend to sleep, and treat it as a floor rather than a ceiling. The practical reason: that is roughly how long it takes the displaced hour to stop being displaced. A boundary at 10 PM for an 11 PM bedtime means the feed ends before the last hour — the hour that would have been spent on it goes instead to the things that actually end the day.
The other half of the default is the easier-to-keep version of the rule: rather than “stop using the phone,” which is a rule about a device, make it “the feeds close at a set time,” which is a rule about the specific things that keep you up. The phone can still do its jobs — the alarm, the audio, the message — inside the last hour; it is the streams that need the earlier boundary. A rule about the feeds is one you can live with; a rule about the device is one you will spend the evening negotiating.
What matters more than the exact number
Three things predict whether the change helps more than the cutoff you pick. Whether the boundary is repeatable — a time you can keep every night rather than an aspiration you meet twice. Whether the last stretch is filled — the hour needs a book or a bath or a quiet thing, not just an absence. And whether the feed is the thing being removed — a calm, finite use of the phone is a different problem than the scroll, and the rule that targets the scroll specifically is the one that works.
Public sleep guidance adds the point that makes all of this matter: regular sleep and wake times are among the most consistently recommended habits for sleep health — which means the boundary's real job is protecting a consistent bedtime from the thing that keeps pushing it later. The 30 or 60 minutes is just the mechanism; the consistency is the goal.
Sources: Healthy Sleep Habits (NHLBI)
When the number matters less than the pattern
A closing qualification that is worth more than the specific advice: if the scroll is not the thing pushing your bedtime — if your sleep is fine and the phone is a calm, finite part of the evening — then the precise cutoff is a question you do not need to answer. The guidance matters most for the pattern it is aimed at: the feed that runs to the pillow, the bedtime that keeps sliding. If that is not your night, the honest answer to “how long before bed” is “you may not need one” — which is the rarest and most under-said answer in this whole genre.
The honest answer to give someone
So the real answer to “how long before bed” is: long enough that the scroll cannot take the last hour. For most people that is somewhere in the 30-to-60-minute range before the intended sleep time — and the intended time matters more than the abstract recommendation. A boundary that protects a bedtime you actually keep is worth more than a more impressive cutoff you do not.
And the version that makes it easy is worth saying plainly: this is exactly what a fixed nightly block does. 9to9 closes the feeds at 9 PM — which for most people is more than an hour before sleep — and the reason that works is not the specific hour but the fixedness: the boundary exists, every night, before the tired part of the evening starts negotiating with it.
What to do with the hour you get back
A closing thought the timing question usually misses: the point of putting the phone down earlier is not the gap itself — it is what fills it. An empty hour before bed is a harder hour than a scrolled one, which is why the recommendation fails when it is just a rule about the phone. The 30-to-60 minutes need a job: the wind-down, the book, the bath, the quiet thing that ends the day. The boundary creates the space; the space needs a use, or the phone will find its way back into it.
The number is less important than the fact of a number. Pick one, make it repeatable, fill the hour it creates — and let the bedtime stop being the thing the feed keeps borrowing from.