White noise machines are one of the most recommended baby-sleep products on the internet, and also one of the least regulated. There’s no dB limit printed on the box, no standard test for “maximum volume,” and — despite what a lot of parenting blogs imply — no single AAP number that every machine is legally required to meet. What actually exists is a peer-reviewed measurement study, a set of occupational-noise reference numbers borrowed from adult workplace safety, and a 2023 AAP policy statement that raises the concern without prescribing an exact dB ceiling. Here’s what each of those sources actually says.
What the 2014 study measured
The number most “is white noise safe for babies” articles are built on traces back to one paper: Hugh et al., “Infant Sleep Machines and Hazardous Sound Pressure Levels,” published in Pediatrics in 2014. Researchers at the Hospital for Sick Children in Toronto bought 14 commercially available infant sleep machines, ran each at maximum volume, and measured the sound pressure level at three distances — 30 cm, 100 cm, and 200 cm (roughly 1, 3.3, and 6.6 feet) — applying a correction factor for a 6-month-old’s smaller ear canal.
Two findings from that paper get cited constantly, and both check out against the published abstract:
- At 30 cm, every one of the 14 machines exceeded 50 A-weighted decibels (dBA) at max volume — the noise ceiling commonly used in hospital newborn nurseries.
- Three of the 14 machines exceeded 85 dBA at that distance — the level NIOSH treats as hazardous for adults over an 8-hour workday (more on that below).
The study’s own recommendation wasn’t a specific number of feet — it was practical placement advice: keep the machine as far from the infant as possible, never inside the crib or clipped to the crib rail, run it at the lowest volume that’s still useful, and don’t leave it running at full volume all night. A lot of the “AAP says 7 feet” language you’ll see online is a rounding of that 200 cm test distance, not a verbatim AAP rule — the study tested at that distance, it didn’t mandate it.
A second, more recent look at the category — a 2024 review in Sleep Medicine that tested 24 machines and 6 phone white-noise apps, reported by ABC News — found every device exceeded the NIOSH 85 dBA/8-hour occupational limit at some setting, and the study’s authors suggested the safe ceiling for infants is probably closer to 60 dBA, not 85. We haven’t been able to locate the full primary text of that study to confirm its methodology independently, so treat the 60 dBA figure as a researcher’s stated opinion rather than an adopted standard — it hasn’t been formally endorsed by the AAP or NIOSH.
What the AAP actually says (and doesn’t)
The AAP’s most current, specific statement on noise and children is “Preventing Excessive Noise Exposure in Infants, Children, and Adolescents” (Pediatrics, November 2023), a policy statement covering noise exposure broadly — concerts, headphones, toys, and sleep machines together, not sleep machines as their own category. On infant sleep machines specifically, the statement is notably restrained: it says pediatricians should counsel parents on safe use and flags that “while some studies show potential benefits in the use of infant sleep machines, one study raised potential concern about sound levels from these devices” — a direct reference to Hugh et al. It does not publish its own dB-at-distance table for sleep machines the way it does for, say, safe sleep positioning.
What the statement does give is a general framework worth applying to sleep machines even though it wasn’t written just for them:
- Dose matters as much as volume. As the AAP puts it, noise exposure is a function of duration and level together — turn one up, and the other needs to come down.
- A practical loudness check: if an adult has to raise their voice to talk to someone at arm’s length, the environment is too loud for a child.
- Hospital newborn nurseries are held to an average of 50 dBA over any one-hour period — the benchmark most “keep it under 50 dB” advice for home sleep machines borrows from, even though that standard was written for a clinical environment, not a nursery at home.
So the honest summary is: there’s no single AAP-mandated decibel number for home sleep machines, but there is a converging, defensible target — ~50 dBA, measured at crib distance, for as short a duration as the baby actually needs it — built from the hospital nursery standard, the 2014 study’s own findings, and the AAP’s general dose-based framework.
The NIOSH 85 dB reference everyone borrows
None of this is really about infant hearing research on its own — the “85 dB” number that keeps showing up is NIOSH’s recommended exposure limit (REL) for occupational noise: 85 dBA averaged over an 8-hour shift, the threshold at which U.S. workplaces are required to run a hearing-loss prevention program. NIOSH’s exchange rate is aggressive: for every 3 dBA increase, the “safe” exposure time is cut in half — so 88 dBA is only safe (by that standard) for 4 hours, 91 dBA for 2 hours, and so on.
That standard was written for adults doing 8-hour shifts, not infants sleeping 10–12 hours a night next to a machine parked a foot from their head — which is exactly the researchers’ point in both the 2014 and 2024 papers. Borrowing an adult occupational limit as an infant safety ceiling is already a conservative comparison; a machine that fails the adult 8-hour standard at crib distance is a clear signal to turn it down or move it, whatever the “correct” infant-specific number eventually turns out to be.
How to check your own machine
You don’t need lab equipment to get a reasonable read on what your machine is actually putting out at crib distance:
- Use a real sound-level-meter app, not just any “dB meter.” NIOSH publishes its own free Sound Level Meter app for iOS, built and lab-tested against a reference meter — accurate to roughly ±2 dBA using the phone’s built-in mic, and closer to ±1 dB with an external calibrated microphone. It’s a reasonable free stand-in for a real meter; most novelty “decibel meter” apps in the app stores have no published accuracy testing behind them and can be off by a wide margin, especially cheap phone microphones at the low and high ends of the range.
- Measure at the distance the baby’s head will actually be, not right next to the speaker grille — the 2014 study’s own numbers make clear how fast sound falls off with distance, so where you put the machine matters more than which model you buy.
- Measure at the volume you actually use, not the maximum — the headline “3 of 14 exceeded 85 dB” numbers from these studies are all at full volume; most machines are quieter than their worst-case spec once turned down.
- Recheck after any firmware or app update if you use a connected machine — volume curves and default levels can change between versions.
Machine features that actually matter
Marketing copy leans on sound-library size and app polish; the features that matter for the safety question are narrower:
- A volume limiter or capped maximum. Some connected machines (Hatch’s Rest line, for instance, publishes decibel guidance in its own support documentation) let you cap the top of the volume range in software, which removes the “someone bumps it to max at 2 a.m.” failure mode entirely. Independent testing has found the uncapped maximum output on some of these devices well above the 50 dBA nursery benchmark, so the cap is doing real work — set it, don’t rely on the default.
- Non-looping audio. Short looped clips (a few seconds repeated) can have audible seams that are more startling than continuous sound; longer or generated (non-looping) tracks avoid the click or jump at the loop point.
- A timer / auto-shutoff. Since duration is half of the dose equation per the AAP’s own framework, a machine that shuts off after 30–90 minutes once the baby is asleep — rather than running at full volume all night — is doing meaningful work on the “dose” side even if the peak volume is identical to a machine without a timer.
- A stated max-volume spec, if the maker publishes one. Very few do. Independent measurements (not manufacturer marketing copy) are the more reliable number when available — for reference, published third-party testing has put a fan-based machine like the Yogasleep/Marpac Dohm in the roughly 50–63 dBA range across its speed settings, while some digital machines with a full 1–10 or 1–100 volume scale (LectroFan is a commonly cited example) can reach the high 70s to mid-80s dBA at their loudest setting — which is exactly the range the 2014 and 2024 studies flagged as a problem at crib distance.
Using white noise as an adult
The safety numbers above are about infants specifically — smaller ear canals, all-night exposure, no ability to move away from the sound or say it’s too loud. Adult use, most commonly for general sleep or to mask tinnitus, is a different calculation.
For tinnitus, sound generators and white-noise machines are a recognized part of “sound therapy,” listed by the American Tinnitus Association as a standard management option. The evidence for how well it works is genuinely mixed: a Cochrane-style review of sound therapy trials for tinnitus covering six trials and 553 participants found no strong evidence that masking alone reduces tinnitus loudness or severity, though it also found no evidence of harm — sound-generating devices appear safe to use. The review’s practical caution is worth repeating: if you have to turn the volume up to an uncomfortable level to cover the ringing, you’re not a great candidate for straight masking, and it’s worth talking to an audiologist about alternatives (like sound enrichment at a lower, non-masking volume) rather than just turning the machine up.
For general adult sleep, the same physics apply as for infants, just with more margin: an adult can wake up and turn a machine down, and typically isn’t sleeping with it a foot from their head all night. The same NIOSH 85 dBA/8-hour reference is the relevant ceiling if you run a machine for a full night’s sleep at close range — most adults use white noise well under that, closer to soft-conversation level (40–50 dBA), which is also roughly where sound-therapy guidance for tinnitus tends to land.
The bottom line
There’s real, peer-reviewed measurement data behind the “some sleep machines are too loud” concern — it isn’t a manufactured scare. Fourteen machines tested at max volume in 2014, all over 50 dBA at 30 cm, three over 85 dBA; a 2024 follow-up across 30 devices found the same pattern. What the AAP adds on top isn’t a specific number — it’s a framework: volume and duration trade off against each other, hospital nurseries hold to 50 dBA average, and a machine’s maximum volume tells you almost nothing about what it’s actually putting out at the setting and distance you actually use. Measure your own setup once with a real meter app, place the machine away from the crib rather than in it, cap the volume if the machine lets you, and use a timer instead of running it all night at full volume.