A Maine elementary school has been running a fire watch since September 2. A power problem on the peninsula knocked out both the fire alarm and the door-card system at Boothbay Regional Elementary School, and the panel still isn’t fixed. Teams of contracted guards have walked the halls in rotating three-hour shifts for most of a month, checking every room while the district waits on replacement parts.
Here’s the direct answer: when a fire alarm system is out of service in an occupied building for more than four hours in a 24-hour period, the common adopted fire and life safety code requires the owner to either evacuate the building or place a fire watch until the system is repaired. NFPA 72 governs the alarm system and the impairment process itself, but the evacuate-or-watch decision comes from the adopted building and fire code your local authority having jurisdiction (AHJ) enforces — and the AHJ decides the details for your building.
A dead panel, a district that stayed open, and a fire marshal’s decision
On September 2, a pair of power issues in Boothbay hit both the elementary school’s fire alarm panel and its door-access system at the same time. The alarm remained armed but unreliable. Rather than close the school, the district consulted the Maine Fire Marshal’s Office, which agreed the building could stay open on one condition: a documented fire watch running continuously until the panel was replaced.
The district contracted a security firm on September 4 to replace the fire panel, smoke detectors, and pull stations. Parts delays pushed the installation date to early October — a month after the outage began. In the meantime, rotating teams of vetted guards have walked every room on three-hour shifts, in addition to normal school staff, watching for smoke and fire the dead panel can no longer detect on its own.
This is what makes an alarm impairment different from almost any other building emergency. The hazard isn’t visible. Nothing is on fire. The building looks completely normal. The only thing missing is the system that would tell anyone if that changed — and the code treats that gap as serious enough to require a human stand-in the moment it runs past a set window.
NFPA 72 defines the impairment; the adopted code sets the fire watch clock
NFPA 72, the National Fire Alarm and Signaling Code, governs how alarm systems are designed, installed, tested, and maintained. It requires a documented process when a system is impaired: someone has to be notified, the monitoring company has to know the signal isn’t reliable, and the system has to be tagged so nobody assumes it’s live. NFPA 72 treats tagging as standard guidance, though many local jurisdictions require it outright.
What NFPA 72 does not do is order a fire watch. That decision belongs to the building and fire code your jurisdiction has adopted, applied by the AHJ. In practice the two work together: NFPA 72 defines when a system counts as impaired, and the adopted code defines what the building must do about it. The widely cited rule — evacuate the building or place a fire watch once an occupied alarm has been down more than four hours in a 24-hour period — comes from that adopted code layer, not from NFPA 72 itself.
Local adoption changes the details. Florida applies fire watch rules through the Florida Fire Prevention Code, adopted by State Fire Marshal rule under FS Chapter 633, pulling in NFPA 1 and NFPA 101 rather than the International Fire Code. Wherever your building sits, the four-hour figure is a common starting point, not a universal statute — your AHJ interprets the four-hour clock for your specific occupancy and system.
A partial outage, a planned test, and an emergency failure each trigger the rule differently
Not every alarm impairment looks like Boothbay’s. The clock and the scope both shift depending on why the system went down and how much of the building it covers.
A planned outage — testing, panel replacement, sensor swaps — still requires notifying the AHJ and the monitoring company ahead of time, and a fire watch is often required if the building will be occupied overnight during the work. An emergency outage, like a power failure or water intrusion that takes the panel offline without warning, requires immediate notification and typically triggers a fire watch right away, running until the system is verified back online. A partial outage — one zone, one floor, one wing — leaves the scope up to the AHJ and the property’s insurer; they may require a watch only over the affected zone rather than the entire building, but that decision needs to be in writing, not assumed.
Boothbay’s case combines a total-panel emergency outage with an extended repair timeline. That combination is exactly why the fire watch has run for weeks rather than hours: the fix depends on parts delivery, not on anyone’s schedule, and the AHJ’s approval to stay open was conditioned on the watch continuing for as long as the panel stays down.
A fire watch replaces automatic detection with a person, a radio, and a log
When the alarm is down, the building loses its automatic detection layer. The fire watch exists to fill that specific gap — not to replace occupants’ normal fire drills, not to guard the property generally, but to detect fire the dead panel can no longer catch on its own.
On a functioning fire watch, guards conduct continuous rounds that cover every area the impaired system used to serve. They carry a working phone or radio at all times, because the notification chain that would normally run through the alarm panel now runs through them. If they find smoke, heat, or an ignition source, they notify the fire department and building occupants directly, following the site’s emergency plan rather than waiting on a signal that isn’t coming. Every round gets a timestamped log entry — who patrolled, when, and what they found — because that log is the only record proving continuous coverage existed during the outage.
The guard’s authority during an impairment is specific, not general. Detection and notification are the job. Firefighting, code enforcement, and building operations decisions stay with the fire department and the property team. A guard who finds a hazard calls it in and clears the area; they don’t attempt suppression beyond a portable extinguisher on a small, controllable fire, and the plan should say so explicitly before the watch starts.
The four-hour clock counts cumulative minutes, not a single continuous stretch
Property managers often assume the four-hour threshold only applies to one uninterrupted outage. That’s not how most adopted codes read it. The clock is typically cumulative within a rolling 24-hour period, which means three separate 90-minute gaps in one day can trip the same threshold as one 4.5-hour outage. A system that keeps faulting and resetting throughout a shift can quietly cross the line without anyone noticing until the AHJ asks for the outage log.
That’s one more reason the notification call to the AHJ belongs at the start of the outage, not after four hours have already passed. An AHJ who knows about an intermittent fault from the first fault can help scope the watch correctly. An AHJ who finds out after the fact is far more likely to require the stricter of the two options — evacuation — rather than accept a watch that started late.
What a written fire watch plan needs before the first patrol starts
NFPA 601 and general fire watch practice both point to the same baseline: a documented plan beats a verbal understanding every time an insurer or AHJ reviews the file after the fact. Before the first guard walks the first round, the plan should specify:
- The exact boundary of the area covered — the whole building, or only the zone the dead panel served.
- Patrol frequency and route, timed so every space gets checked on a fixed interval rather than “as time allows.”
- The notification chain: who the guard calls first if smoke or fire is found, and the backup contact if that call doesn’t connect.
- What equipment each guard carries — radio or phone, flashlight, a portable extinguisher rated for the occupancy’s hazards.
- The log format: time, location, observation, and signature for every round, matching what NFPA 72’s impairment process and the AHJ’s order both expect to see.
- The restoration trigger — who tests the repaired system and who has authority to stand the watch down once it passes.
A watch that runs for a single evening still needs this plan; a watch that runs for a month, like Boothbay’s, depends on it. Guard rotations change, shift supervisors change, and the only thing keeping the coverage consistent across four weeks of three-hour shifts is a plan that doesn’t depend on any one person’s memory.
Guard count and shift length usually drive the cost of an extended watch
Property managers calling about an open-ended outage almost always ask the same question first: what is this going to cost while we wait for parts? The rough formula that governs almost every fire watch quote is guard count multiplied by hours of coverage multiplied by the hourly rate, and for an alarm outage the guard count depends entirely on how large an area the dead system served.
A single-zone impairment in a small building might need one guard walking continuous rounds. A full-panel outage across a multi-wing school or a multi-floor office building typically needs enough guards to keep every zone on its patrol interval without gaps — which is why Boothbay is running rotating teams rather than one person covering the whole building alone. Scheduled, planned outages for testing or panel replacement usually price lower than an emergency call placed after hours, because emergency response compresses mobilization time into a window measured in hours, not days.
The other cost variable is duration, and it’s the one owners underestimate most. A watch quoted for an overnight test is a different budget line than a watch that has to run for weeks because replacement parts are backordered. Getting a written proposal early — one that breaks out guard count, shift structure, and an estimated timeline tied to the actual repair schedule — keeps an extended outage from becoming an open-ended expense with no visibility.
An alarm outage triggers a parallel duty to warn employees, not just to detect fire
Fire detection isn’t the only gap an alarm outage opens. Under OSHA’s alarm and detection rules, employers have a separate duty to make sure employees can still be warned if the automatic system that normally does that job is down. OSHA 1910.165 requires a backup means of alerting employees during an outage — runners and telephones are the rule’s own named examples — plus prompt restoration of the system. That obligation runs in parallel with the fire code’s evacuate-or-watch decision; one doesn’t substitute for the other. A property running a fire watch for the AHJ still needs its own answer for how it alerts staff if the guard spots a fire, and a phone tree or runner system only satisfies OSHA if it can genuinely reach every employee fast enough to act.
That’s the layer property managers most often miss during an extended outage like Boothbay’s. The fire watch covers detection. A separate, documented interim alerting plan covers the warning duty. Running both, and keeping records of both, closes the actual gap the dead panel opened rather than half of it.
Documentation the AHJ and your insurer will both ask for
An extended impairment produces a paper trail whether or not anyone plans for one. Building it deliberately from day one is faster than reconstructing it later for an insurer or a fire marshal’s file review.
At minimum, keep the time the system went down and who was notified, the monitoring company’s record that it was told the signal was unreliable, a timestamped patrol log for every single round the fire watch completed, and the exact restoration time along with who verified the new system was live and tested. If an AHJ order or written approval to remain occupied was issued — as it was in Boothbay — keep that document too. A verbal “you’re fine to stay open” from a fire marshal’s office is worth far less than the same statement in writing.
What a property manager should do the moment the panel goes dark
The operational decision at the start of an outage isn’t complicated, but it has to happen fast. Call the AHJ before deciding anything else, because the fire marshal’s office — not the building owner — decides whether the property can remain occupied and what the fire watch has to cover. Notify the monitoring company in the same call so its records match the AHJ’s. Then start a fire watch immediately if the building will be occupied past the four-hour window, rather than waiting to see if repairs land quickly; a watch that starts an hour late still needs a log explaining that hour.
From there, the clock that matters is the repair clock, not the watch clock. A fire watch can run for a single evening or, as Boothbay’s has, for close to a month. What ends it is the same thing every time: the alarm system restored, tested, and confirmed working by whoever the AHJ requires to make that call — not a guess that the parts probably arrived.
Getting a qualified fire watch on site fast is often the difference between a school, warehouse, or office building staying open during an outage and having to send everyone home. The Fast Fire Watch Company runs firefighter-led fire watch guards nationwide, with certified guards on site in under 3 hours, 24/7. If an alarm panel just went dark at your property, call 1-800-899-7524 or visit fastfirewatchguards.com to get a documented fire watch started before the four-hour window closes.
