Carbon monoxide on boats: why half the victims were outdoors, and what a detector can't do
The space under a swim platform can hold six times the concentration that's immediately dangerous to life. A compliant marine CO alarm may take fifteen minutes to sound at a fraction of that. Both facts are true, and together they explain who the detector actually protects.
Most gear writing about carbon monoxide on boats ends at “fit a detector.” That advice is correct and it is not enough, because a large share of the people it would have to save were never inside the cabin.
If you want the buying decision rather than the argument, our guide do you need a carbon monoxide detector on a boat? answers it directly. This piece is the evidence underneath it.
How much carbon monoxide is actually behind a boat?
Enough to kill in minutes. Investigating houseboat poisonings on Lake Powell, the CDC measured 6,000 to 30,000 ppm in the cavity beneath the stern deck with the generators running, up to 7,200 ppm on and near the swim platform, and 200 ppm at water level ten feet away (MMWR 49(49)). Oxygen in that stern cavity fell as low as 12%.
Set those against the exposure limits in the NIOSH Pocket Guide: the immediately dangerous to life or health level is 1,200 ppm. The recommended exposure limit is 35 ppm as an eight-hour average, with a 200 ppm ceiling.
The swim platform reading is six times IDLH. The stern cavity is twenty-five times it.
This is not only a houseboat problem. NIOSH instrumented ten express cruisers under real operating conditions and recorded peaks above 1,000 ppm at the stern on more than one boat. Their conclusion, verbatim: “When the canvas is deployed and boat is underway, CO concentrations exceeded the immediately dangerous to life and health (IDLH) level near the swim platform for many of the evaluated boats.”
Why does exhaust come back aboard at all?
Because a moving boat drags a low-pressure pocket behind its own transom. NIOSH describes “negative pressures at the rear of the boat and cockpit areas, which are generated when the boat is underway, [which] draw exhaust gases, including CO, from behind the boat into inhabited areas.” Boaters call it the station wagon effect.
The conditions that make it worst are counterintuitive: slow speed, into the wind, with the canvas fully up and no forward hatch open. NIOSH measured how high the recirculating exhaust reached above the cockpit floor at different speeds — about 4 ft at 5–10 mph, 3 ft at 15 mph, 2 ft at 25 mph. Going slowly does not make it safer. It makes it worse, and it puts the contaminated layer at the height of a seated adult or a standing child.
Three other paths deliver the same gas:
- Generator exhaust at anchor or rafted up. The CDC’s 6,000–30,000 ppm figures were measured with generators running, not underway.
- Teak surfing and platform dragging — holding the swim platform or transom to body-surf the wake. California made it illegal on 1 January 2005. It puts a person’s head directly in the highest concentration measured anywhere on the boat.
- The boat next to you. ABYC’s own educational bulletin warns that “operation, mooring, and anchoring in an area where other boats’ engines or generators are running may put your boat in an atmosphere containing CO.” No published safe separation distance exists.
Why is CO poisoning so hard to notice on a boat?
Because every early symptom is something you’d expect to feel on a boat anyway. The CDC puts it plainly: “Because symptoms of CO poisoning resemble those of other common conditions (e.g., alcohol consumption, motion sickness, heat stress, and nonspecific viral illness), poisonings often go unrecognized.”
California’s Division of Boating and Waterways says it more concretely: symptoms “may include nausea, dizziness, confusion, headache and fainting; however, people often mistakenly attribute these symptoms to too much alcohol, sun, and noise, or to motion sickness from the water or exhaustion.”
Headache, nausea, dizziness and confusion on a hot afternoon at anchor is a description of a normal day out. That is the entire problem. The Coast Guard notes the same overlap with seasickness and intoxication, “so those affected may not receive the medical attention they need.”
Who does a marine CO alarm actually protect?
People asleep in a cabin. Not a swimmer behind the transom — and the reason is in the certification, not in any particular product.
ABYC A-24 — “Installation of Carbon Monoxide Detectors and Alarms,” July 2020 — requires detectors on boats with an enclosed accommodation compartment, monitoring the contiguous cabin space plus each separate sleeping space. It requires testing to UL 2034, marking as a “Marine Carbon Monoxide Alarm”, and a printed replace-by date. ABYC recommends compliance for boats built after 31 July 2021.
A-24 sets no ppm thresholds itself. It delegates that to UL 2034, and this is the number that reframes everything. From the CPSC’s conformance testing report:
| Concentration | Permitted time to alarm |
|---|---|
| 70 ppm | 60–240 minutes |
| 150 ppm | 10–50 minutes |
| 400 ppm | 4–15 minutes |
And no alarm at all below 30 ppm.
A fully compliant alarm may take four hours to sound at 70 ppm and fifteen minutes at 400 ppm. That is not a defect. UL 2034 is built around how carboxyhemoglobin accumulates in blood over time, which is the right model for someone sleeping in a bunk while a slow leak fills the cabin.
It is the wrong model entirely for 7,200 ppm under a swim platform. At that concentration the exposure is decided long before any compliant device is designed to respond. The alarm is a cabin-and-sleeping safeguard. It is not protection for anyone in the water at the stern.
Say that out loud on your own boat, because the marketing does not.
What does the CDC data say about where people were standing?
Outdoors, mostly. This is the statistic that should change how the subject is taught.
In the CDC’s series, 111 boat-related CO poisonings were identified between 1990 and 2000, nine of them fatal. 74 of the 111 involved houseboats — and 37 of those 74, exactly half, were outdoor exposures. Half of those resulted in loss of consciousness. Autopsy carboxyhemoglobin in the fatal cases ran 49% to 59%.
Open air, on the back of a boat, in daylight.
Current federal counts look small by comparison: the 2024 Recreational Boating Statistics record 5 CO incidents, 2 deaths and 26 injuries. Read that carefully — it counts only reportable accidents, and a poisoning that ends in a hospital visit rather than a damaged vessel often isn’t one. (The same report lists CO as a contributing factor in zero incidents on a different table, which tells you something about how consistently this gets classified.)
What are the detectors themselves specified to do?
Fireboy-Xintex is the name that turns up most on US boats, and the CMD6-M series is the current generation. Published specifications, from authorized dealer listings — Fisheries Supply carries the full sheet:
- Alarm point specified as 10% carboxyhemoglobin, not a raw ppm figure, using time-weighted-average sensing
- Approximately 7 year end-of-life on the sensor
- 12/24 VDC or sealed battery variants, 4 mA draw, 85 dBA horn
- −40 °F to +158 °F operating range
- Conforms to UL 2034 Edition 4 Marine; meets ABYC A-24
- An internal relay for generator shutdown, and interconnection between units
That last one is the feature worth paying for. Generator exhaust at anchor is the dominant at-anchor hazard, and a detector wired to shut the generator down acts while everyone aboard is still arguing about whether they feel funny.
Two notes. The alarm-point-as-%COHb framing is the same reasoning as UL 2034 — it is modeling the person, not the air. And the ~7 year sensor life is specific to this generation; older Fireboy units were rated 3 to 5 years, and blending the two figures gets you the wrong replacement date.
Fireboy has been in Grand Rapids, Michigan since 1973, with a UK operation in Poole. Its core business is fire suppression; CO detection sits alongside it.
Is a household CO alarm good enough?
No, and the cleanest argument is regulatory rather than technical. A-24 requires UL 2034 testing and marking as a “Marine Carbon Monoxide Alarm.” A house alarm is UL 2034 listed but not marine-marked, so it does not satisfy the standard your surveyor or insurer will reference.
The engineering case points the same way — marine units publish operating ranges like −40 °F to +158 °F and run off the boat’s DC bus rather than a 9 V cell — but the certification argument is the one you can check on a label in ten seconds.
What actually reduces the risk?
The detector is the last line, not the first. In order of how much they matter:
Move the exhaust. After the Lake Powell deaths, the National Park Service required concession rental houseboats to fit vertical stack exhaust extending at least nine feet above the upper deck to ABYC P-1, and to convert generators to catalyzed low-CO units. This is the intervention that worked, and it worked by design rather than by alarm.
Keep people off the back of the boat. No swimming, sitting, hanging or dragging at the transom or swim platform while an engine or generator is running. This is the single behavior that the CDC’s outdoor-exposure statistic is about.
Open the front, not just the back. The station wagon effect is a pressure problem. Forward hatches and windows break the recirculation; more canvas at slow speed makes it worse.
Watch the neighbors. Rafting up next to a boat running a generator puts you in their exhaust with none of your own instruments implicating them.
Then fit the alarm, wire it to the generator if you can, and diary the replace-by date — because an expired electrochemical sensor is a plastic box that reassures you.
And if two people aboard develop a headache at the same time, that is not the sun. Get everyone into fresh air first and work out what it was afterwards.
Full write-up on the manufacturer is on the Fireboy-Xintex brand page. We use and recommend Fireboy; nobody paid for this, and when they become a stockist we’ll say so on the page. Nothing here is a substitute for a marine survey or your own equipment’s manual.