A quiet carbon monoxide detector does not mean there is no carbon monoxide in your home.
The alarms sold in most US stores are required to stay silent at 30 ppm, more than three times the level EPA uses as its 8-hour health standard for outdoor air. They are built to catch a poisoning in progress, not to report what you breathe every day.
That is not a defect, and it is not a loophole. It is the safety standard working exactly as it was written, and with the heating season about to start, it is worth knowing what the silence does and does not tell you.
The standard behind the alarm on your ceiling
UL 2034, the Standard for Safety for Carbon Monoxide Alarms, is the standard most retail CO alarms in the United States are built to. CPSC also recognizes the International Approval Services and Canadian Standards Association standards.
For its FY 2012 testing program, staff at the US Consumer Product Safety Commission collected 102 alarms from a single retail source, six models from three manufacturers, and ran 72 of them through the sensitivity test. Matthew Brookman of the Directorate for Laboratory Sciences published the results in March 2014.
The report carries its own caveat on the cover: it “was prepared by the CPSC staff, and it has not been reviewed or approved by, and may not necessarily reflect the views of, the Commission.” Everything below is staff work.
UL 2034 does not simply ask an alarm to detect carbon monoxide. It asks the alarm to sound inside a window, and to say nothing outside it.
At 70 ppm, the alarm must sound between 60 and 240 minutes.
At 150 ppm, between 10 and 50 minutes.
At 400 ppm, between 4 and 15 minutes.
Sounding too early is a failure, exactly like sounding too late. Two alarms failed the 70 ppm test for going off at 33 minutes.
Below that window, the standard requires silence outright. In the report’s own words:
The standard also provides test methods in which the alarms are exposed to 30 ppm CO for 30 days or 8 hours, depending on the particular test. During this period, the alarms must not activate.
CPSC staff ran the 8-hour version, not the 30-day one. The report is explicit that “the 30 day test is to be conducted for 8 hours.” The 30-day method lives in the standard; it was not exercised here.
Why it is built that way
CPSC staff state the reason plainly. The criteria exist:
to prevent premature activation, which may lead to a lack of consumer confidence in the performance of the alarms. A reduction in the perceived reliability of CO alarms could lead to consumers disregarding the alarm entirely, even when dangerous levels of CO were present and detected.
That is a defensible trade. An alarm nobody believes is worse than no alarm at all. But it does mean the device on your hallway ceiling is built to catch a poisoning in progress, not to describe the air you breathe every day.
The two numbers, side by side
EPA’s national health standard for carbon monoxide in outdoor air is 9 ppm averaged over 8 hours, and 35 ppm over 1 hour, neither to be exceeded more than once a year.
The level a UL-listed alarm is required to ignore is 30 ppm: more than three times the 8-hour figure, and just under the 1-hour one.
These are different instruments for different jobs. One is an outdoor ambient air quality regulation, the other an indoor consumer product trigger. Neither is breaking a rule, and nothing here is an exceedance of anything. Set them next to each other anyway, because the comparison is the point: the alarm’s quiet zone reaches well past the level outdoor air is managed to over a working day.
The display is not a readout
This is the part most people get wrong, and it is written directly into the standard. UL 2034 Section 36.8 requires a concentration display to be accurate within plus or minus 30 percent of the indicated amount, and then adds:
No indication shall be given for CO concentrations less than 30 ppm.
A digital CO alarm showing zero is not telling you the air is clean. It is telling you the air is under 30 ppm. CPSC staff noted one alarm that appeared to be programmed not to display anything below 50 ppm at all.
How long the silence actually ran
CPSC staff held 72 alarms at 70 ppm and recorded when each one sounded. Seventy-one eventually did. One never activated.
Median time to alarm: 80 minutes, including the two units that failed for sounding too early. The slowest unit that still passed took 168.5 minutes.
The median and the range are our arithmetic on the report’s Tables 4, 9 and 14. The individual activation times are the report’s own.
One alarm did sound at 30 ppm
The standard requires silence at 30 ppm. That is not the same as every unit delivering it. In the 120 °F high-temperature series, Sample Set E produced “one failure during the 30 ppm, 8-hour nuisance alarm test with a full alarm signal.” A second unit, from Sample Set F, gave a trouble signal. Those were the only two nuisance-alarm failures recorded across both temperature conditions. Note the condition: this was not a room-temperature result.
The limits, stated plainly
This was a test chamber, not a house. Nobody was exposed to carbon monoxide and no health outcome of any kind was measured. The report does not show that 30 ppm in a home harms anyone, and this article does not claim it does.
The alarms were manufactured in 2011 and 2012 and tested in 2012. Current models may perform differently. All 102 came from a single retail source, which is a real sampling limitation.
CPSC staff also flag a limitation in their own equipment. Air moved across the alarms in the chamber at nearly 300 ft/min, against roughly 20 ft/min in a typical home and in UL’s own test chambers, and staff wrote that the difference “may influence their performance and skew the results.”
What survives all of that is narrow and solid. A silent CO alarm is evidence that carbon monoxide is below the alarm’s threshold. It is not a measurement of your air.
Why this matters in September
CDC counted 2,244 unintentional carbon monoxide deaths in the United States between 2010 and 2015, with the highest numbers every year falling in winter months. Of the 393 deaths in 2015, 36 percent came in December, January or February.
The furnace that has been off since spring is about to run.
What to do
Service the furnace, water heater and any gas appliance before the heating season, not during it.
Put an alarm in the hallway near the bedrooms in each separate sleeping area. That is CPSC’s own placement guidance.
Check the manufacture date. These units carry an end-of-life signal for a reason.
Read a zero on the display as “under 30 ppm,” never as “none.”
If flu-like symptoms turn up in more than one person in the house, or ease when you leave, do not wait for the alarm. That is general carbon monoxide safety guidance rather than a finding of this report. Get out, get fresh air, and call 911 or your gas utility from outside.
Bottom line
A carbon monoxide alarm is an evacuation trigger, not an air quality monitor. It was designed that way deliberately, and it is good at the job it actually has. Asking it to also tell you whether your air is clean is asking it to do something it was specifically built not to do.
If you want to know what is actually in the air in your house, that is a measurement, and it is a different instrument.
illumenair measures indoor air in Portland and Vancouver homes.
971-363-5626 · info@illumenair.com · www.illumenair.com
References
Brookman MJ. Carbon Monoxide Alarm Conformance Testing to UL 2034: Standard for Safety for Carbon Monoxide Alarms. Phase I, FY 2012 Test Results. US Consumer Product Safety Commission, March 2014.
US EPA. NAAQS Table. Updated November 2025.
CDC. QuickStats: Number of Deaths Resulting from Unintentional Carbon Monoxide Poisoning, by Month and Year, United States, 2010-2015. MMWR Morb Mortal Wkly Rep 2017;66(8):234.