Every carbon monoxide alarm sold in the United States is loud on paper. The number printed in the manual is almost always the same one: 85 decibels at 10 feet. It is a reassuring figure, it appears in specification sheets without context, and it is the only acoustic fact most buyers ever hear about the device they are hanging outside a bedroom.
The problem with that number is not that it is wrong. The problem is that it is measured under conditions that have very little to do with the condition that matters most — a person asleep, behind a door, some distance away, at three in the morning. Loudness at ten feet is an engineering test result, not a guarantee that anyone heard it.
This guide walks through where the 85 dB requirement comes from, what happens to it as it travels through a house, why the standards community has already acknowledged that high-frequency tones underperform for sleeping occupants, and what that means for how you should think about a CO alarm's role in a layered setup.
The Short Answer: 85 dB at 10 Feet
UL 2034, the standard covering single- and multiple-station carbon monoxide alarms, requires the audible alarm signal to produce 85 dB minimum at 10 feet (3 meters). You will find that sentence, nearly word-for-word, in the manual of essentially every residential CO alarm on the market.
Two clarifications matter immediately:
- "Minimum" runs in the right direction. 85 dB is a floor, not a typical value. A unit can be louder and still comply — most are marginally louder at the horn itself.
- 85 dB is measured at the alarm, not at the pillow. The measurement distance is ten feet from the device in open air. It says nothing about what arrives at a sleeper's ear two rooms away.
There is one important relaxation built into the installation codes. When an alarm is intended to notify occupants in the same room where it is installed — a unit mounted inside a bedroom, for example — the required sound pressure level drops to 75 dBA at 10 feet. That lower figure appears in state-level guidance derived from NFPA 720 and is the basis for the common "one alarm inside each bedroom" recommendation.
So the honest short answer is: loud enough to meet a standard written around proximity, and silent in the specification about everything beyond it.
Where the Number Comes From
The 85 dB requirement is a product standard, not a building-code acoustic target. It lives in UL 2034 as a performance test on the device itself. The codes then reference the standard and add installation language on top.
The layered structure looks like this:
| Layer | What it governs | What it says about sound |
|---|---|---|
| UL 2034 | The alarm product | Audible signal ≥ 85 dB at 10 ft |
| NFPA 720 / NFPA 72 | Installation and placement | Alarms near sleeping rooms must be clearly audible in all bedrooms over background noise with intervening doors closed |
| State and local codes | Enforcement | Usually adopt the above; vary on bedroom-level requirements |
The critical phrase in the installation layer is "clearly audible in all bedrooms over background noise levels and with all intervening doors closed." That is a performance requirement placed on the *installation*, not on the device. The device passed its 85 dB test on a bench. Whether your hallway unit satisfies "clearly audible behind a closed door" is a property of your floor plan, your door, and the frequency of the tone — none of which UL 2034 evaluates.
This is the gap where most real-world disappointment lives. For a fuller walk through what the standard does and does not cover, see our explainer on the UL 2034 standard.
What 85 Decibels Actually Sounds Like
Decibels are logarithmic, and most people's intuition about them is poor. A reference scale helps:
| Sound | Approximate level |
|---|---|
| Whisper at 1 m | 30 dB |
| Normal conversation | 60 dB |
| Vacuum cleaner | 70 dB |
| Busy street traffic | 80–85 dB |
| CO alarm at 10 ft (UL 2034 minimum) | 85 dB |
| Motorcycle at 25 ft | 95 dB |
| Nightclub | 110 dB |
Two things are worth pulling out of that table.
First, an alarm at 85 dB sitting ten feet away is genuinely loud — comparable to standing next to moderately busy traffic. If you are awake, in the room, and not wearing headphones, you will hear it. There is no scenario in which a compliant CO alarm is quiet up close.
Second, notice that the values in the middle of the table are not far apart. 85 dB is only about 15 dB above ordinary conversation. Because the scale is logarithmic, a modest drop in delivered sound pressure moves you a long way down that table — from "busy street" toward "vacuum cleaner" and below. And unlike the street or the vacuum, an alarm competes with a house at night that has its own noise floor: a running fan, a humidifier, a television in another room, traffic outside.
The Ten-Foot Rule, and Why Distance Eats the Number
In open air, sound pressure falls roughly 6 dB for every doubling of distance from the source. This is the inverse-square law, and it applies to alarm horns closely enough to be useful for planning:
| Distance from alarm | Approximate level from an 85 dB source |
|---|---|
| 10 ft (the rating point) | 85 dB |
| 20 ft | ≈ 79 dB |
| 40 ft | ≈ 73 dB |
Those are free-field figures with nothing in the way. A house is not free field.
Now apply it to a common layout. A single CO alarm is mounted in the hallway outside two bedrooms. The hallway unit is 85 dB at ten feet — but the second bedroom's pillow is 25 feet away and around a corner. By distance alone the arriving signal is already down near 77 dB. Add a closed hollow-core door, and the delivered level at the pillow can land in the 50s or low 60s — below the level of the vacuum cleaner in the table above, in a room where the occupant is asleep.
This is precisely why placement guidance keeps converging on the same answer: an alarm inside or immediately outside each sleeping room, not one per floor. Our guide to how many CO detectors you actually need walks the arithmetic room by room, and the placement height guide covers the vertical dimension that most people get wrong.
The Frequency Problem: High Tones Do Not Travel Through Doors
Loudness is only half the acoustic story. The other half is pitch, and it is the half almost nobody talks about.
The audible signal in a conventional residential alarm is generated by a piezoelectric horn. Historically these sit around 3,000–3,100 Hz — a high, piercing tone. High-frequency sound is easy to generate cheaply and it cuts through mid-range background noise well. It also has a property that matters enormously here: high frequencies are attenuated far more than low frequencies by walls and closed doors.
The fire-protection research community has known this for a long time. Work by Butler, Bowyer and Kew in the early 1980s found that lower-frequency alarm tones transmit through a closed door measurably better than traditional high-frequency tones. That finding stayed in the literature for two decades before it changed a code.
What changed it was a separate body of sleep research. Studies by Dorothy Bruck and Ian Thomas, conducted under the auspices of the Fire Protection Research Foundation, compared waking effectiveness across population groups and found that a 520 Hz low-frequency square-wave signal waked high-risk groups — people over 65, people with hearing loss, school-age children, and people impaired by alcohol or sleeping medication — substantially better than the conventional 3,100 Hz tone. For those groups, the low-frequency signal achieved equivalent waking at sound pressure levels as much as 20 dB lower.
That research is why NFPA 72 now specifies a low-frequency audible signal with a fundamental frequency of 520 Hz ± 10 percent for notification in sleeping areas where governing codes require it, and why the 2021 International Fire Code picked up low-frequency requirements for sleeping rooms in new hotels, dormitories and apartment buildings.
Read that carefully, though, because the scope is narrower than it sounds. The NFPA 72 residential chapter does not currently require a low-frequency signal in every sleeping room of a single-family home — it requires it where a hearing impairment is being accommodated. Low-frequency residential alarms are not yet common in the consumer market. The committee records on this are explicit that cost-benefit work was still incomplete when the requirement was written.
Which leaves the ordinary buyer in an awkward spot: the research says the tone your alarm makes is not the optimal tone for waking a sleeping person, and the market largely still sells you the old tone.
What the Manufacturers Put in Their Own Manuals
This is the part that surprises people, because it is printed in the same manuals that advertise the 85 dB figure. Language of this kind appears across major CO alarm brands:
> "CO Alarms may not be heard. The alarm horn loudness meets or exceeds current UL standards of 85 dB at 10 feet. However, if the CO Alarm is installed outside the bedroom, it may not wake up a sound sleeper or one who has recently used drugs or has been drinking alcoholic beverages."
> "This device is not intended to alert hearing impaired residents."
> "CO Alarms may not waken all individuals."
And the mitigation printed alongside those warnings is not a better alarm. It is a drill:
> "It is recommended that you hold a fire drill while family members are sleeping in order to determine their response to the sound of the CO Alarm while sleeping and to determine whether they may need assistance in the event of an emergency."
That instruction is worth taking literally. If you have never tested whether a specific sleeping person wakes to your specific alarm in your specific house, you do not actually know the answer — you know the decibel rating of a horn on a bench.
Note also what is missing from these manuals. There is no decibel figure for "behind a closed door, two rooms away." The standard does not require one, so nobody publishes one.
Sleep, Age, and Who the Tone Was Never Calibrated For
Human arousal threshold during sleep is not a fixed number. It varies with sleep stage, age, and individual hearing — and it is considerably higher than most people assume.
The groups identified in the waking-effectiveness research as poorly served by a conventional 3,100 Hz tone are not exotic edge cases:
- Adults over 65. Arousal threshold rises with age, and so does the prevalence of high-frequency hearing loss, which is exactly the band the piezo horn occupies.
- School-age children. Children have higher arousal thresholds than adults for high-frequency signals; this was one of the more counterintuitive findings in the literature, since parents routinely assume the opposite.
- Anyone with hearing loss. See the numbers below.
- Anyone who has consumed alcohol or taken sedatives. Both raise arousal threshold materially.
The scale of the hearing-loss group alone is large enough that it cannot be treated as a niche. Per the National Institute on Deafness and Other Communication Disorders, approximately 15% of American adults — about 37.5 million people — report some trouble hearing, and roughly one in eight people aged 12 or older (13%, or 30 million) has hearing loss in both ears by standard audiometric examination. About 28.8 million U.S. adults could benefit from hearing aids, and among those aged 70 and older who could benefit, fewer than one in three has ever used one.
Untreated hearing loss is the common case, not the rare one. And the hearing that goes first with age is high-frequency hearing — again, the band the alarm occupies.
The Structural Gap: an Alarm Is a Threshold Event
Step back from acoustics for a moment and look at what an audible alarm actually is as an information channel.
An alarm is a binary, threshold-triggered, single-location, time-limited broadcast.
- Binary. It has two states: silent or sounding. Nothing in between is communicated.
- Threshold-triggered. It speaks only after conditions have already crossed a defined point. Below that point it is indistinguishable from a unit that is not there.
- Single-location. It reports on the air at its own sensor and nowhere else.
- Time-limited. It communicates during the event, and to a person who must then interpret it correctly while asleep.
Every one of those properties is a reasonable engineering trade-off for a device whose job is to be a last line of defence. None of them makes an alarm a good instrument for *knowing what is happening*.
That is also why a push notification cannot stand in for the sounder. An alert produces zero decibels at the pillow; it arrives on a phone that may be on silent, in another room, or face-down on a nightstand. The network layer is covered in our guide to Wi-Fi carbon monoxide detectors — the short version is that connectivity is a courtesy copy, never the primary notification.
That distinction matters because the questions people actually ask are not threshold questions. "Is the furnace venting correctly this morning?" "Did the level in the garage go up after I ran the car?" "Is that reading from the water heater or from the oven?" These are questions about a continuous value at a specific place, and an alarm cannot answer any of them. It can only tell you, eventually, that a line was crossed.
This is the same distinction drawn in our comparison of smoke detectors and carbon monoxide detectors — different threat, same structural limitation: a siren is a decision device, not a measurement device.
What a Second Channel Adds
If the acoustic channel has known failure modes — distance, doors, frequency, sleep stage, hearing — the rational response is not to abandon it. It is to stop making it the *only* channel.
A visual readout on a monitor changes the shape of the information in four specific ways:
- It is continuous rather than threshold-triggered. A display shows the current reading, including readings well below any alarm point. You can observe a trend before it becomes an event.
- It is spatially flexible. A monitor can sit where the question is — beside the mechanical room, on the garage-side wall, at the top of the basement stairs — rather than only where code requires an alarm.
- It does not depend on hearing or on being awake. A number on a screen is available to a person who slept through the tone, and to a household member with hearing loss.
- It is inspectable. You can check it deliberately, on your schedule, without waiting for the device to decide to speak.
None of this replaces a code-required alarm, and no responsible setup treats it as a substitute. UL 2034 alarms remain the device that is actually required, and the one that sounds. A supplemental monitor is a second layer that answers a different question: not "has a threshold been crossed" but "what is the air doing right now."
For households where the acoustic gap is widest — a bedroom at the far end of a hallway, a sleeper who has never been tested against the alarm, an occupant over 65, a home with closed doors at night — the second layer is where the practical answer lives.
If you want the full picture on what a display does and does not tell you, our guide to CO detectors with a digital display covers the readout side in detail.
Adding that channel does not require rewiring the house. The VASAIKO HD16 is a plug-in monitor with a live ppm readout that keeps sensing on its internal rechargeable cell when mains power drops — the audible layer stays exactly as it is, and a second one sits alongside it.
How to Test Whether Your Alarm Actually Reaches You
You can measure this in an evening, without buying anything beyond a free smartphone app.
1. Test with doors in their night position. Close every bedroom door, turn off fans you would normally run, and press the test button from the hallway alarm. Walk into each bedroom and listen from the pillow position. An app will give you a rough A-weighted level; treat it as comparative rather than calibrated, since phone microphones are not measurement instruments.
2. Compare rooms, not absolutes. What you are looking for is the *spread*. If the hallway reads very loud and the far bedroom is marginal, that gap is your problem — not the absolute number.
3. Run one genuine sleep drill. Pick a morning you do not need to be sharp, and have someone trigger the test from the hallway while you are asleep. People are frequently surprised by the result, in both directions.
4. Check the frequency assumption. If anyone in the household is over 65, has diagnosed hearing loss, or is a school-age child, treat a marginal result as a failure rather than a borderline pass. That is exactly the population the waking-effectiveness research identified.
5. Re-test after any layout change. Moving furniture, replacing a hollow-core door with a solid one, or adding a fan changes the delivered level.
If the answer from any bedroom is marginal, the fix is not a louder alarm — 85 dB is already the floor and there is not much headroom above it. The fix is a unit closer to the sleeper, a display you can check, or both. A monitor with a live readout covers the second half of that sentence without touching the first.
Where the Monitor Goes
Placement for a supplemental monitor follows the question you are asking, not the code minimum.
- The mechanical boundary. Next to the furnace, boiler or water heater, where incomplete combustion would first show up. See our guides on furnace and boiler CO detection and water heater CO risk.
- The garage-side wall. The shared surface between an attached garage and living space is where a car left running shows up first. Covered in attached garage CO alarms and garage CO detection.
- The basement top-of-stairs. Where a lower-level appliance's output migrates upward. See basement CO alarms.
- The sleeping level. Where the acoustic channel is weakest and the occupants are least able to respond.
The VASAIKO HD16 is built for this second layer: plug-in, with an internal rechargeable cell so it keeps sensing through an outage, and a screen that shows the live CO reading alongside combustible gas, smoke, temperature and humidity. It is a supplemental monitor, not a replacement for the alarms your code requires, and it is tested to UL 2034, 1484 and 217 standards rather than positioned as a code-compliance device.
Frequently Asked Questions
How loud is a carbon monoxide alarm, exactly?
UL 2034 requires a minimum of 85 dB measured at 10 feet. Where an alarm is installed inside the room it is meant to notify, installation guidance derived from NFPA 720 permits 75 dBA at 10 feet.
Is 85 dB loud enough to wake you up?
It depends entirely on distance, doors, and who is sleeping. Manufacturers' own manuals state that an alarm installed outside a bedroom may not wake a sound sleeper, and that CO alarms may not wake all individuals.
Does a CO alarm with a digital display sound louder than one without?
No. The audible requirement is the same 85 dB at 10 feet regardless of whether the unit has a screen. The display adds a second, non-audible channel — it does not make the horn louder.
Why does my alarm sound quieter in the bedroom than in the hallway?
Distance costs roughly 6 dB per doubling, and a closed door attenuates high-frequency tones significantly. A hallway alarm rated 85 dB at ten feet can arrive at a distant pillow well below conversational level.
What frequency is a CO alarm tone?
Conventional residential alarms use a high-frequency piezoelectric tone around 3,000–3,100 Hz. Research on waking effectiveness has favored a 520 Hz low-frequency signal for sleeping areas, particularly for older adults, children and people with hearing loss.
Are low-frequency CO alarms required in homes?
Not generally. NFPA 72 specifies a 520 Hz signal for sleeping-area notification where governing codes require it — typically where a hearing impairment is being accommodated — and low-frequency residential units are not yet common in the consumer market.
What should I do if someone in the house has hearing loss?
Treat the audible channel as unreliable for that person and add a non-audible one. Roughly 15% of American adults report some trouble hearing, so this is a mainstream case, not a special one. Our digital display guide covers what a visual channel provides.
How do I know if my alarm is loud enough without special equipment?
Close the doors as they are at night, press and hold the test button, and listen from each pillow position. Then do one real sleep drill. Comparative results between rooms matter more than any single number.
Does the alarm get quieter as the unit ages?
The horn itself does not meaningfully weaken, but the sensor does drift over a service life of roughly five to ten years, and a drifting sensor alarms later rather than quieter. See how long CO detectors last.
Is a plug-in alarm louder than a battery one?
No — the audible requirement is identical. What differs is power continuity: a plug-in unit with a battery backup keeps sensing through an outage. Covered in plug-in versus battery detectors.
What is the difference between the alarm pattern and a low-battery chirp?
A CO alarm is a repeating pattern — typically four quick beeps, a pause, then repeat. A low-battery chirp is a single short beep every 30–60 seconds. Our guide to CO detector beeping and false alarm causes break both down.
Does a louder alarm mean better protection?
Not necessarily. Loudness is already standardized at a floor; the variables that actually decide whether anyone hears it are placement, distance, door state and the sleeper. Adding a visible reading addresses those variables in a way a louder horn cannot.
The Bottom Line
85 dB at 10 feet is a real and useful number, and it is worth knowing that it measures a horn on a bench rather than a household at night. Between the alarm and a sleeping person sit distance, doors, a high-frequency tone that doors absorb, and a waking threshold that rises with age, hearing loss, alcohol and sleep stage. The standards community has already documented most of this; the market simply has not caught up.
The practical response is not to hunt for a louder alarm. It is to stop relying on a single channel. Put code-required alarms where they belong, put one inside or immediately outside each sleeping room, run a sleep drill to find out what your house actually delivers — then add a second layer you can read rather than only hear.
The VASAIKO HD16 is designed for that second layer: a plug-in supplemental monitor with a live CO readout and rechargeable standby power. If you are mapping out the rest of your home, start with how many detectors you need and where to put them, and if you want the ppm numbers behind the thresholds, the safe CO ppm levels chart lays them out.