Bottom line up front: the working target is around 50 A-weighted decibels at your baby’s ear, with the machine as far from the crib as the room allows — and the number nobody checks is duration, because risk scales with level and time together. Most sound machines can exceed that target easily at close range, almost none of them tell you where their volume dial actually sits, and a phone or a $20 meter settles the question in five minutes.
I measure this because the alternative is guessing, and every machine in my house read higher than I expected. Affiliate links never change my verdicts.
The two numbers that define the limit
Two figures are worth memorizing, and they do different jobs.
50 dBA is the nursery reference. It is the recommended sound limit for infants in hospital nurseries, and the benchmark the sound-machine research is measured against. Not a safety cliff — a conservative environmental target for a baby sleeping in that room for months. Aim here.
85 dBA over eight hours is the exposure ceiling. NIOSH sets its recommended exposure limit at 85 dBA averaged over an eight-hour day, and — this is the part that matters for overnight sound — for every 3 dBA increase, it recommends halving the exposure time. That is an adult occupational standard, not a pediatric one — the outer bound of the argument rather than a target. The fact that a nursery device can reach it at all is why this post exists.
Every sound machine on the market lives in the gap between those two numbers.
What the machines actually output
The reference study is a 2014 paper in Pediatrics that measured 14 infant sleep machines at maximum volume, at three distances, corrected for a six-month-old’s ear canal. The Children’s Hospital of Philadelphia’s write-up summarizes the headline result: every machine tested exceeded 50 decibels, and several exceeded 85.
The mean maximum levels by distance are the useful part:
| Distance from machine | Mean maximum output |
|---|---|
| 30 cm — on the crib rail | around 79 dBA |
| 100 cm — beside the crib | around 71 dBA |
| 200 cm — across a small room | around 63 dBA |
Three of the fourteen exceeded 85 dBA at 30 cm. Read that as: a machine on the crib rail at full volume can put an infant into adult occupational-exposure territory for a whole night. Nobody buys one intending that, and it is achievable by accident.
Distance is the free volume control
That table is a physics lesson. Sound pressure falls with distance — roughly 6 dB per doubling in open air, less in a small room with soft furnishings bouncing it back. Moving from crib rail to far wall bought about 16 dB in that study, a bigger reduction than most people get from the volume dial.
So the order of operations is: move it first, then turn it down. A machine across the room at a medium setting will almost always beat the same machine at a low setting on the rail. Guidance on how far varies more than it should. The study’s authors recommended placing the device as far from the infant as possible and never in the crib or on the rail; the CHOP write-up cites more than 200 cm; some manufacturers say three feet. When sources disagree, take the most conservative one that your room physically allows.
Which published specs are real, and which are marketing
Almost no product page publishes a decibel figure. The Yogasleep Dohm product page lists two fan speeds and a tone dial and no output number at all. You have to go to the company’s own support FAQ to learn that the fan-based Dohm units run roughly 50–75 dB, that the portable units span zero to about 85 dB, and that its loudest machine reaches about 90 dB. Useful disclosure, published where no shopper will look.
Room-conditional claims are not device claims. Hatch’s support documentation answers the volume question by saying its maximum sits below 50 decibels at five to ten feet in an average carpeted nursery, and notes that carpet alone can cut the level substantially. Every word of that is probably true and none of it describes the device — it describes a room. On a hard floor, at three feet, the same machine is a different measurement. I like the hardware on other grounds, as my Hatch Rest review goes into, but this is the claim shape to notice.
Treat these as meaningless until a number appears: “baby-safe volume”, “gentle for newborn ears”, “pediatrician-designed”, and any auto-limiting claim without a stated dB figure and distance. No consumer standard forces a sound machine to declare its output, which is why the study’s authors asked manufacturers to limit levels and add warnings.
What to buy for control rather than loudness. The most useful hardware feature is fine-grained volume steps at the quiet end. A unit in the LectroFan mold gives you many small increments, which lets you land on 50 rather than choosing between 43 and 61. A two-speed mechanical machine cannot, so you adjust distance instead — workable, less precise. Full testing notes are in the white noise machine roundup.
For how long: all night, or off after the fall-asleep?
This is the contested part, and I would rather show the disagreement than pretend it is settled.
The 2014 study’s authors recommended running a machine only briefly, until the infant is asleep. A lot of pediatric sleep practice goes the other way, because a sound environment that changes between sleep cycles is a common cause of a 45-minute nap — and Betteroo’s rundown of what sound machines are for sits on that side.
The resolution is the 3 dB exchange rate: duration is only a problem in combination with level. All night at a genuinely low level across the room is a different exposure from all night at maximum on the rail. If you want eleven hours, the price is measuring rather than guessing; if you would rather not measure, a timer is the safer default. Either way, audit volume creep — the dial only ever travels one direction, and my post on weaning off white noise is mostly about weaning the volume rather than the habit.
The five-minute audit
- Put a phone flat where your baby’s head lies, with a free sound-level app open, and run the machine from its usual spot at its usual setting.
- Read the average, not the peak. Phone microphones are uncalibrated and can be off by several decibels — approximate is still enough to catch a 70.
- If it reads high, move the machine before you touch the dial. Re-measure.
- For a real number, a basic sound level meter — the RISEPRO HT-80A type, around $20, 30–130 dB with A-weighting — beats a phone and also settles a noisy street-facing room.
- Write the setting down. In three months somebody will have turned it up.
FAQ
How many decibels should white noise be for a baby?
Around 50 dBA at the baby’s ear is the commonly cited target, mirroring the recommended limit for hospital nurseries. Levels near 85 dBA are where adult occupational exposure limits begin, and a machine at full volume on a crib rail can get there.
How far should a sound machine be from the crib?
As far as the room allows, and never in the crib or on the rail. More than 200 cm — about seven feet — is the figure the pediatric write-ups use; some manufacturers suggest three feet, which is the least conservative advice in circulation. Distance is the most effective control you have.
Can white noise damage a baby’s hearing?
At sensible levels and distances, there is no evidence it does. The concern in the research is specific: machines run near maximum volume, at close range, for many hours a night, reaching levels that would be capped in an adult workplace. That combination is avoidable, and avoiding it costs nothing.
Is it safe to run white noise all night?
It depends on the level. Low volume across the room all night is a different exposure from high volume at the crib. If you cannot measure, use a timer; if you measure and land near 50 dBA, continuous sound is reasonable and avoids the sound environment changing mid-nap.
