A custodian at an Antioch, California school district walks the halls and counts the panels beeping and flashing red. He estimates 80 percent of the district’s fire alarm systems are not working right. The district had recently cut the electronics technician position that used to keep those panels maintained. When the local fire department sent an inspector out, they confirmed enough of the problem to place two campuses, including Black Diamond Middle School, on formal fire watch status.
Yes, a malfunctioning fire alarm at an occupied building like a school can trigger a mandatory fire watch, and it usually does once the local fire department confirms the system cannot reliably notify anyone. The trigger is not the blinking light on the panel. It is whether the system can still do its actual job: detect a fire and get word to the fire department fast enough to matter. When that function is gone, NFPA 72 and the adopted fire code hand the building two options, and “wait for the vendor” is not one of them.
In the Antioch case, the fire captain overseeing the response drew a sharp distinction that property managers and school administrators should hold onto: the part of the system that would alert people already inside the building was still working. The part that had failed was the communication link to the fire department itself. That single distinction decided whether a fire watch was required, and it is the same distinction that decides it at your property.
A fire watch after an alarm failure depends on which function actually broke, not how bad the panel looks
A flashing trouble light does not automatically mean evacuate or staff a watch. What matters is which piece of the alarm’s job is broken: local notification (horns and strobes that alert people already in the building), or off-site notification (the signal that reaches the fire department, directly or through a monitoring company). The Antioch schools kept local notification. They lost the connection to the fire department. That is exactly the gap a fire watch exists to fill.
Under NFPA 72’s fire alarm impairment framework, the code that governs alarm systems does not itself order a fire watch. NFPA 72 defines the impairment and requires a documented process: notify the people responsible for the building, tag the system as out of service, and get it repaired. The decision to evacuate or staff a watch comes from the adopted fire and life safety code working alongside NFPA 72, and it is the local authority having jurisdiction (AHJ) who applies that rule to your specific building.
Occupied buildings commonly get a 4-hour clock once the alarm can’t reach the fire department
The rule most fire marshals cite is straightforward: an occupied building with its alarm system out of service for more than 4 hours in a 24-hour period must be evacuated or placed under an approved fire watch, with the AHJ notified either way. That threshold is common under widely adopted fire and life safety codes, not a single uniform national law. Local adoption and interpretation vary, and the AHJ can move faster than the clock if conditions warrant it.
Two things about that clock matter more than the number itself. First, it starts at the moment of impairment, not at the moment someone finally reports it. A custodian noticing a beeping panel for weeks before anyone escalates it is exactly the scenario that turns a manageable impairment into an enforcement problem — the clock has been running the whole time, whether or not anyone was watching it. Second, an alarm panel that still sounds locally but cannot reach the fire department is not “mostly working.” For code purposes, off-site notification failure is treated the same as a dead panel, because the fire department cannot respond to a fire it never hears about.
Multi-building campuses add a wrinkle worth naming directly. A school district with 20-plus sites and one contractor covering all of them is not one impairment — it is potentially 20 separate ones, each running its own clock from the moment its own system failed. Contra Costa County’s fire captain noted the department was aware of some pre-existing issues but not the full scope, which is exactly the failure mode a documented, per-building impairment log is supposed to prevent.
Losing the fire department connection is the specific failure that forces the decision
Central-station monitoring, when a building has it, only reports signals from a system that is actually working. When the panel can no longer reach the fire department — whether the monitoring line failed, a technician position that maintained the system was eliminated, or a component broke — the building has lost the one function no amount of local horn-and-strobe activity replaces: someone outside the building finding out fast enough to respond.
That is the exact condition NFPA 72 and the adopted fire code treat as requiring immediate action. It doesn’t matter that people inside the building would still hear an alarm if one went off. If the fire department has no way to be told, the building has no automatic path to a response, and a human being has to become that path until the system is fixed.
The impairment math is simple, and a facilities team should be running it before the AHJ does
The formula a fire marshal is quietly checking is: hours since off-site notification failed − the allowed window (commonly 4 hours) = hours of exposure with no compliant plan. Anything above zero gets worse, not better, the longer it sits, because it means the building has been relying entirely on luck for however many hours the panel has been unable to call for help.
Run the Antioch numbers as an illustration. If a campus’s alarm lost its connection to the fire department at, say, 7 a.m. on a school day, the common 4-hour window would run out by 11 a.m. — while students, teachers, and staff are still on site for the rest of the school day. A district running that math across 20 buildings, each with its own failure date, is not managing one impairment; it is managing a portfolio of them, and the exposure calculation has to run separately for each site until every one is confirmed restored or watched.
This is also why “we’ve had these issues for a while” — which is close to what the Antioch custodian described — is the worst version of this math to discover after the fact. If a panel has been beeping for weeks and nobody escalated it to the AHJ or arranged interim coverage, the exposure number is not hours; it is potentially hundreds of hours across the affected buildings. That is the gap between an impairment that was managed and one that becomes a finding.
Documentation is what turns “we knew about it” into a defensible response
An AHJ or an insurer reviewing an alarm impairment after the fact is not grading whether the building ever had a problem. Almost every building eventually does. They are grading whether the response was documented as it happened. The record a fire marshal or a district’s insurance carrier will ask to see typically includes:
- The date and time the impairment was first identified, and by whom.
- Whether the failure affected local notification, off-site notification to the fire department, or both — confirmed by the alarm contractor or the fire department, not assumed.
- The written notification to the AHJ, including what is impaired and the proposed interim plan.
- A timestamped patrol log for every fire watch round, covering every building the impairment reached.
- The date and method of restoration, and who verified the system was actually reconnected to the fire department — not just that the panel stopped beeping.
A district or property that can produce all five of those items has a defensible record even if the underlying equipment failure was embarrassing. A district that can only say “we knew the panels were acting up” does not, and that gap is exactly what turns a maintenance failure into a compliance finding once an inspector or a reporter starts asking questions.
A fire watch guard covering an alarm outage has four jobs, not one
Once a watch is ordered, a guard standing near the front office is not a compliant fire watch. The role has four specific functions, and the Antioch order — continuous patrol of the affected campus until communications are repaired — spells out exactly why all four matter for a facility this size.
Patrol. The guard covers every area the impaired system served, on continuous rounds through the shift, not a single walk-through at the start. On a multi-building campus, that means every wing, not just the office where the panel lives.
Detect. With the fire department connection down, the guard is standing in for the notification function specifically — watching for smoke, unusual odors, and anything the panel would have flagged if its signal were getting out.
Notify. The guard has to know the manual path to the fire department: a phone call, a radio, whatever the interim plan specifies. A guard who spots a hazard but has no direct way to reach the fire department has not closed the gap; they have only moved where the failure happens.
Log. Every round gets a timestamped entry — what was checked, what was found, who was told. This is the record the AHJ reviews to decide whether the watch can be lifted once the system is verified restored. An unlogged round is treated as one that did not happen.
On OSHA’s alarm and detection rules (1910.164 and 1910.165), the employer’s duty to alert people does not pause just because the fire code side is also handling the impairment. 1910.165 specifically requires a backup means of alerting occupants while a system is down — the rule names runners and telephones as examples — and that duty runs alongside, not instead of, the fire watch the fire code requires for the detection-and-notification gap. A school running both a physical watch and a manual alert plan is not being redundant; it is covering two separate legal requirements that happen to point at the same underlying failure.
Schools and multi-building campuses face a version of this problem larger properties should plan for
The Antioch case is a school district story, but the underlying pattern — a shared maintenance function gets cut or delayed, and the impairment surfaces slowly across many buildings instead of all at once — applies to any organization managing more than one site under one facilities budget. Hotels with multiple towers, hospital campuses with separate wings on separate alarm loops, and multi-building industrial sites all share the same exposure: a single point of failure in staffing or contracting can leave several buildings impaired at once without anyone tracking each one’s individual clock.
The fix is not complicated, but it has to be deliberate. Every building or zone on a shared alarm contract needs its own impairment record, checked independently, because the AHJ and any insurer reviewing the response afterward will not accept “the contractor was handling it” as a substitute for a per-building log. Antioch’s district ultimately did contract fire watch coverage for the affected sites — the lesson is arranging that before an inspector forces the question, not after.
Property and facilities managers have three decisions to make before the next shift starts
The hours after discovering an alarm impairment — whether it is one building or twenty — are when the clock is already running and the decisions get made correctly or not.
First, confirm exactly what failed. “The panel is beeping” is not a diagnosis. Get a specific answer: is local notification intact, is off-site notification to the fire department down, or both? That distinction alone determines whether you are looking at a fire watch or a more limited response.
Second, notify the AHJ and propose an interim plan in writing. “Approved fire watch” means approved by the fire code official, not self-certified by the property or the district. Waiting for the fire department to find the problem on their own, the way Antioch effectively did, is the version of this that ends up in the news.
Third, staff the watch to the scope the AHJ actually accepts, and treat every additional building or zone discovered later as its own decision. A facilities team managing multiple campuses under one contractor cannot assume a fix at one site clears the others. Each one needs its own confirmation that the fire department connection is restored before the watch on that building comes down.
Skipping any of these steps does not make the requirement disappear. It just means the fire watch requirement gets discovered by an inspector or a reporter instead of by the property’s own planning — usually with less time to arrange staffing and less credibility with the AHJ going forward.
An alarm system that can no longer reach the fire department is not a maintenance backlog item. It is a building operating without the one function that gets help moving before a small problem becomes a large one, and the fire code treats it that way. Confirm what actually failed, call the fire marshal before an inspector calls you, and get a guard on site with a log running for every building still on the list.
The Fast Fire Watch Company provides nationwide fire watch coverage for schools, districts, and commercial properties working through an alarm system impairment or an AHJ-ordered watch. Guards are trained on the detection, notification, and logging duties discussed above, and every shift comes with a timestamped patrol record your fire marshal can review on request. Review our California fire watch requirements page for state-specific guidance, or call 1-800-899-7524 for certified guards on site in under 3 hours, 24/7.
