White noise for newborns and babies: what's proven, how loud, how long

It's eleven at night. The phone sits on the shelf next to the crib with the white noise on, and it will stay that way until morning. If that's what you do too, you're not alone: in a survey from a pediatric ENT clinic in the United States, published in 2026, 59% of parents of children up to age eight used white noise, and 91% of them left it on all night, on average nearly ten hours a day.

That exact part, all night, at home, with a healthy baby, is the part no one has tested in a controlled study. What has been measured is something else, and it's worth knowing before you pick a volume: a controlled study in newborns that followed them for five minutes, and two sets of measurements on machines and apps that show how loud they can get. It's all below, with the numbers and with what they don't say.

The short answer: white noise helped newborns fall asleep faster. In the 1990 randomized trial in healthy newborns, after five minutes 80% of those with the noise were asleep, against 25% of those without. That it makes them sleep longer at night is not proven. If you use it: outside the crib, starting from the lowest volume, under 50 dBA at your baby's head, and switched off once they're asleep; if you leave it on all night, at the lowest volume that does the job.

What's proven: falling asleep, within minutes

In 1990, at a London hospital, 40 healthy newborns between two and seven days old were randomly split into two groups. A white noise machine was switched on in the cribs of the first group; the others stayed in silence. After five minutes, 16 of 20 were asleep with the noise (80%) and 5 of 20 without it (25%).

The result is real, and it's cited everywhere. The rest of the study is worth reading too:

• It measured falling asleep, not sleep. It recorded only whether the baby was asleep after five minutes. How long they slept afterward, how many times they woke, what the night was like: none of that.
• The machine switched itself off after about four and a half minutes. What the study actually tested was a short sound, not one left on all night.
• It was loud, and it was close. The machine sat in the crib, 30–50 cm from the baby's head, and at 30 cm it measured 72.5 dB for the first 30 seconds, then 67 dB. That's exactly what today's recommendations tell you not to do, and well above the 50-decibel target discussed below.
• It worked only in babies who weren't hungry. The authors say so explicitly. Sound doesn't take the place of a feed.
• A single observer, who knew which group each baby was in, decided who was asleep. And in 36 years no one has repeated the experiment in the same form, with either a positive or a negative result.

Older studies, from the ’60s and ’70s, found the same thing in the short term: a continuous sound calms a newborn more than silence does. Two details from them still matter today. First: heartbeats were nothing special. They didn't work better than a metronome or lullabies, and a rhythm close to the baby's own mother's pulse didn't calm them more than a different one. Second: the effect grew with the volume. A louder sound calmed more.

That leads to a problem that rarely gets mentioned: the effect was seen at volumes higher than the ones considered cautious, and at a cautious volume no one has measured it in healthy babies. That doesn't mean a quiet sound doesn't help. It means we don't know.

What isn't proven: all-night sleep

We looked for the studies that tested white noise at home, in healthy babies, over whole nights. No controlled study has tested it on its own. The closest is a series of four one-year-olds who resisted going to bed and woke at night: white noise at 75 dB, switched on at bedtime and left on all night, after which three of them slept better. No comparison group, data recorded by the parents, and a volume far above the 50-decibel target. Three other studies, randomized ones, put it in a package of techniques alongside swaddling, rocking, “shh” or a pacifier:

• In the United States, 51 mothers of healthy newborns were given either a video of the techniques or one about routine care; 35 finished the study. At 6 and 12 weeks there was no difference, in either crying or sleep.
• In Turkey, 42 mothers learned the techniques face to face, at home. Their babies had fewer wakings and more sleep, but as reported by the mothers themselves, and the mothers in the intervention group knew they had received the training.
• Also in Turkey, 84 first-time mothers were given soothing training in the maternity ward, with a swaddling blanket and a sleep toy that plays white noise. Their babies slept more and cried less, but sleep was only a secondary outcome of a study about the mothers.

The big numbers that get passed around, like “two extra hours of sleep” or “three more hours”, come almost entirely from the neonatal intensive care unit: preemies in incubators, some on breathing support, on a ward where the background noise has nothing to do with your bedroom. They don't apply at home. A 2021 systematic review, across all ages, rated the quality of the evidence that continuous noise improves sleep as “very low”.

For colic, the only randomized trial compared white noise with rocking, in 40 one-month-old babies in Turkey. Crying dropped by half with both, and white noise came out slightly ahead: 8–11 fewer minutes of crying a day. But there was no group that got nothing, and the babies entered the study after a week with nearly five hours of crying a day; after a week that hard, the next one usually turns out easier whatever you do, so that halving can't be put down to the sound. And the sound was on only while the baby was crying, not all night. How evening crying changes, week by week.

How loud: where “under 50 decibels” comes from

Guides for parents usually give a number between 45 and 65 decibels. The most quoted one, 50, has a precise origin, and that origin also tells you what the number is not.

It comes from criteria for hospital newborn nurseries, published in 1999: the sound around the bed should not go above 50 dBA as an hourly average, with separate limits for peaks. The stated aim was to protect the sleep and vital signs of hospitalized babies, preemies among them. Newer standards for intensive care lowered the target to 45.

Three consequences:

• It isn't a hearing threshold. The levels at which hearing damage comes into the discussion are 25–35 decibels higher: 85 dBA over eight hours is the occupational limit for adults, which isn't designed for children, and one proposal for children stops at 75 dBA as a 24-hour average. 50 is a target for sleep, not the edge of a cliff.
• It's an average of all the sounds in the room. A machine running steadily at 50 dBA uses up, on its own, the entire limit designed for a hospital ward.
• It hasn't been tested at home. No one has checked that 50 is safe or that 55 isn't. It's a cautious, borrowed target, and that's how to use it.

For reference: a whisper is about 30 dBA, a quiet home 40, light traffic 50, a normal conversation 60–70. Every 10 extra decibels sounds roughly twice as loud. So 50 dBA is quieter than a normal conversation, not “about as loud as a quiet conversation”, as many places say.

How loud machines and apps get, measured

Two teams have measured how much sound these devices put out. The first, in 2014, took 14 infant sleep machines and turned them all the way up. The averages, adjusted for the ear of a six-month-old: 79.1 dBA at 30 cm, 70.5 at one meter, 63.3 at two meters. All 14 went above 50 at one meter, and 13 of 14 went above 50 even at two meters. Three went above 85 at 30 cm, the level at which, after eight hours, the occupational limit for adults is exceeded.

The second, in 2021, measured 8 machines and 6 phone apps, at minimum and maximum volume, at three distances, with no ear correction, so the numbers can't be compared directly with the ones above. This is the part that concerns you if you use your phone. The averages, in dBA:

Read the table by columns. If you move the phone from 30 cm to one meter, the level drops by 5–8 decibels. If you turn the volume from maximum to minimum, at one meter, it drops by 30–35. The volume button moves several times more decibels than distance does. At maximum volume and 10 cm, meaning the phone placed in the crib, 9 of the 14 devices went above 85 dBA. At minimum volume and one meter, the apps stayed between 28 and 41 dBA, but the loudest machine still reached 53.

“Two meters” often shows up as a rule. The authors of the 2014 study suggest that more than two meters may be safer, but they say just as clearly that they found no safe distance: at maximum volume, 13 of 14 machines went above 50 there too. Distance helps. Volume decides.

How long

The 2014 authors' recommendation has three parts: the machine as far from the baby as possible, never in the crib or on the crib rail; the volume as low as possible; and a limited duration, with a timer or switched off once the baby is asleep. The American Academy of Pediatrics (AAP) tells parents the same thing: as far from the baby's head as possible, and only for a short time.

Duration matters because the only experiments with long-term exposure are in animals. Rat pups kept in continuous white noise at 65–70 dB, roughly what a machine at maximum puts out at one meter, 24 hours a day, from the seventh to the fiftieth day of life, showed delayed maturation of the auditory cortex: the part of the brain that learns to tell sounds apart. The delay went away after two months in a normal environment. Few studies measured hearing itself: in the adult rats it stayed normal, but one study in pups, at 70 dB, found raised hearing thresholds. In another experiment, in adult rats and with a rhythmic noise, ten hours a day had almost the same effect as 24, and the effect lasted at least six weeks after the noise was turned off.

None of this lets you say that white noise damages a baby's hearing or speech: the rats were in noise day and night, for weeks on end, and the species hears differently. But you can't say the opposite either. In babies, there is no study that has followed hearing or speech after months of white noise at night. The two studies in children that we found saw no differences, but one was a survey about parents' concerns, and the other tested the two groups with different sounds. That isn't evidence of safety. It's only the absence of evidence of harm.

If you've used it loud or close to the crib for months, that doesn't mean it harmed your baby: in babies there is no evidence of that. Starting tonight, move it farther away and turn it down, and at the next checkup tell the doctor how you've been using it. The signs that call for a hearing test are the same in every baby, with or without white noise: they don't startle at loud noises; after six months they don't turn toward a sound; by one year they don't say any simple words, like “mama” or “dada”; they turn when they see you, but not when you just call their name; they seem to hear some sounds and not others. If you see any of these, don't wait: ask for a hearing test, even if your baby passed one at birth. It's quick, it doesn't hurt, and it can even be done while they sleep.

That leaves two ways to use it, and it's worth choosing one with your eyes open.

1. Just for falling asleep. You switch it on, they fall asleep, you switch it off, or you set a timer. It's what the authors of the measurements and the AAP recommend, and it's actually what was tested in 1990: a sound that lasted a few minutes.

2. All night. It isn't what the AAP recommends, but there are two arguments for it. One is covering noises you don't control: the neighbors, the streetcar, the birds that wake them at five in the morning. The other is that, when your baby wakes between sleep cycles, they find the same conditions they fell asleep in. Both make sense, and neither has been tested in a controlled study in babies. In adults, in the lab, the results contradict each other: in a small test with four people, white noise added on top of noise recorded in an intensive care unit prevented the micro-arousals the noise caused; in a recent one with 25 adults, pink noise added on top of intermittent noises reduced REM sleep. If you choose this option, the main rule is the lowest volume that does the job, measured under 50 decibels at your baby's head.

White, pink, brown, rain or heartbeat?

The difference between the colors is one of spectrum. White noise has the same energy at all frequencies and sounds like the hiss of a radio between stations. Pink noise puts more energy into the low frequencies, and brown noise even more, so they sound progressively deeper.

In babies, no one has compared the colors with each other. There is no study on brown noise in infants, and so no “right” age for it either. There is none with rain, ocean or forest sounds in healthy babies. And the old comparison between kinds of sound, the one with heartbeats, a metronome and lullabies, did not find heartbeats better than the others.

So choose based on two criteria that actually matter:

• Steady, with no peaks. The hospital criteria limit the average and the peaks separately, and the World Health Organization's guideline for bedrooms sets a separate limit for individual noise events. A hiss or a light rain has no peaks. A thunderstorm, birdsong or a gong does. This is an inference from the noise criteria, not a result tested in babies, but it costs nothing.
• One you can stand. If you leave it on all night, you hear it all night too.

What gets passed around and doesn't hold up

A few claims show up on almost every page about white noise. We checked them at the source:

• “Your newborn spent nine months listening to 80 dB of noise.” A fetus starts to hear at around 19 weeks, and at first only low sounds. In the womb, the level reaches 60–85 dB only below 100 Hz; above 500 Hz it's around 40 dB. The measurements were made with an underwater microphone in 10 women in labor, and the authors called the womb “relatively quiet”.
• “Total silence stresses babies out.” We found no study showing this in a healthy, full-term baby.
• “Mom's heartbeat is the most soothing sound.” In the womb, the mother's heartbeat could be heard in only 2 of the 10 women measured, and in the old studies heartbeats didn't soothe better than other continuous or rhythmic sounds.
• “Ocean waves sync with your baby's breathing.” An 8-second wave means 7.5 cycles per minute. A sleeping newborn breathes 30–60 times a minute, or once every 1–2 seconds. There is no study on anything like this.
• “Brown noise is for babies 3–12 months old.” There is no study on brown noise in babies.
• “Studies confirm it's safe for long-term use.” There is no study that has followed babies' hearing or speech after months of white noise.
• “Pediatricians recommend white noise all night.” The AAP's advice for parents is: as far from the baby's head as possible, and only for a short time.

In short, what to do

• Outside the crib, never on the crib rail. As far away as your room allows, charging cable included: at least one meter from the crib, where your baby can't grab it. The AAP lists dangling cords and electrical wires among the strangulation risks around the crib.
• Start from the lowest volume and turn it up only as far as it takes to cover what you need to cover. Never at maximum, and never with the above-100% boost that some apps have.
• Measure once, before you put your baby down. Start the sound on the phone that will play it, set in its spot, and place a second phone with a sound meter app on the mattress, where your baby's head will be. The target: under 50 dB; then take that phone out of the crib. The number is only a rough guide: on the iPhone that was tested, below 50 dB the apps read higher than the real level, but the authors warn that the result can't be generalized to other phones. If it reads close to 50, turn it down a little more. Measure again when you change the sound, the volume or where the phone sits.
• For falling asleep, switch it off once they're asleep, or set a timer.
• If you leave it on all night, choose a steady sound and the lowest volume that covers the noise you're trying to block.
• Never headphones, or a speaker pressed against their ear. Sound delivered straight into a small ear canal ends up louder than it does in an adult's.
• Check once that you can still hear your baby. If you sleep in another room, make sure the monitor picks up crying over the noise. No one has studied this, but the test takes a minute.
• It doesn't help a hungry or overtired baby. In the 1990 study it worked only in babies who weren't hungry. And if your baby struggles to fall asleep every night, the cause lies somewhere else.

How Lulla helps

Lulla has 50 sounds, among them white and brown noise (“White Noise” and “Brown Noise” in the app), “shh”, heartbeats, rain, ocean, wind, household sounds and a few instruments. The first ten are free, among them rain, which sounds steady, and ocean, which rises and falls with the waves; white noise, brown noise, “shh” and heartbeats are part of Premium. The sound loops until you stop it, and the app has no shut-off timer, so the duration part of this guide is up to you: once your baby is asleep, you switch it off. It also stops if you close Lulla from your recent apps or if you start other music or a video on the same phone. The volume slider goes up to 300%: above 100%, the app boosts the sound digitally, and at 300% it raises it by up to about 9.5 dB over the original level. That's the boost mentioned above, and you don't need it for a baby's sleep: keep the slider at 100% or below. The slider doesn't start at the minimum either: the first time you use it, it sits at 70%, so check it before bedtime. The final volume also depends on the phone's volume buttons, so measure with both set the way you'll leave them overnight.

Where it really helps is with the question no study can settle for your child: does it work for them? Time every sleep in the app, and in the sheet that appears when you stop the timer, check “White noise” under “Sleep conditions” and note the awakenings. Do this for one week with sound and one without, and double-check that box every time, because the app pre-fills the conditions from earlier sleeps. Then compare in History how long each sleep lasted and how many awakenings you noted. With Premium, Stats puts the average hours of sleep per day for the last 7 days next to the 7 days before; awakenings and the sound checkbox don't show up there, so you compare those yourself. It isn't a study, and your baby keeps growing in the meantime, but it's about them. Forty newborns from 1990 can't tell you that.

To keep in mind: this article is informational, not medical advice, and the figures quoted come from the studies in the source list, each with its own limitations. For any concern about your baby's hearing, sleep or development, your pediatrician remains the first port of call: they have the full history.