Commercial guideUpdated July 20268 min read
How often must exit and emergency lights be tested?
Monthly for 30 seconds, annually for 90 minutes, with the records kept in the building. The schedule is the easy part. What catches property managers is which code section applies, what a self-testing fixture does and does not excuse, and the fact that every unit passing does not mean the building passes.
The short answer
Every emergency light unit and every illuminated exit sign in your building gets an activation test monthly, for not less than 30 seconds. Battery-powered units additionally get a power test annually, running on battery for not less than 90 minutes. Written records of both are kept on the premises and produced when the fire code official asks.
That is the whole schedule, and if you do only that you are ahead of most buildings. The rest of this guide is the four things that turn a building doing the testing into a building that passes the inspection, because those are not automatically the same.
Which code actually applies
Search this question and nearly every result cites NFPA 101, the Life Safety Code, section 7.9.3. That guidance is not wrong about the intervals. It is just not the document your building is inspected against.
Buildings in Santa Clara County are inspected against the California Fire Code, Title 24 Part 9. Section 1032.10 states that emergency lighting shall be maintained in accordance with Section 110 and inspected and tested in accordance with Sections 1032.10.1 and 1032.10.2. Section 1032.10.1, the activation test, sets the monthly 30 second interval. Section 1032.10.2, the power test, sets the annual 90 minute run for battery-powered equipment.
The numbers land in the same place, so the schedule you follow is the same either way. The reason to know the difference is what happens after an inspection. A correction notice cites a California Fire Code section, and a printout of NFPA 101 is not a response to it. If you are assembling a compliance binder for a portfolio, cite what your inspector cites.
The self-testing trap
This is the one that costs buildings money twice, first for the fixtures and then at the inspection.
Section 1032.10.1 allows the monthly test to be performed manually or by an automated self-testing and self-diagnostic routine. Property managers read that, buy self-testing fixtures, and reasonably conclude the monthly walk is now handled by the hardware.
The same section continues. Where testing is performed by self-testing and self-diagnostics, a visual inspection of the emergency lighting equipment shall still be conducted monthly to identify any equipment displaying a trouble indicator or that has become damaged or otherwise impaired.
So self-testing does not remove the monthly obligation. It changes the nature of it, from pressing and holding a button at each unit to walking the building and reading indicators. That is genuinely faster and it is a good reason to buy the fixtures. It is not a reason to stop walking the building. The failure mode we find is a corridor where three units have been blinking a fault at nobody for fourteen months.
Passing is not compliant
Every unit can light for the full 90 minutes and the building can still be written up, because the code governs how much light reaches the egress path, not just whether fixtures illuminate.
Section 1008.3.2 sets initial emergency illumination at not less than an average of 1 footcandle, with a minimum of 0.1 footcandle at any point, measured along the path of egress at floor level. It permits that to decline to 0.6 footcandle average and 0.06 footcandle minimum by the end of the duration, which is the code acknowledging that batteries fade. It also caps the maximum-to-minimum uniformity ratio at 40:1.
That last one is the requirement almost nobody tests for. A bright twin-head unit directly over a door, with a long dim stretch of corridor forty feet away, can blow the uniformity ratio while every individual fixture passes its own test. The practical version: during the annual test, walk the route a person would actually take to get out, not from fixture to fixture. Dark gaps between working lights are common, they are cheap to fix once identified, and they are invisible to a fixture-by-fixture check.
On the 90 versus 60 minute confusion
Both numbers appear in the code and they answer different questions. Section 1008.3.1 requires the emergency power system in new construction to provide power for not less than 90 minutes. Section 1104.5.1 provides for 60 minutes in certain existing-building situations. The annual power test itself is written at not less than 90 minutes for battery-powered equipment. Which combination governs a given building depends on its age, its occupancy, and what the local jurisdiction has adopted or amended, so confirm it rather than assuming. Testing to 90 minutes satisfies the more demanding case, which is why it is the number we work to.
The records are half of it
Section 108.3 requires a record of periodic inspections, tests, servicing, and other maintenance to be kept on the premises or another approved location for not less than 3 years, unless a different period is specified elsewhere, and to be made available to the fire code official.
Read that phrase again: on the premises. A log that lives in a management company's office across the county, or in a service contractor's system, is not where the code puts it. When an inspector is standing in your lobby, the answer needs to be in the building.
The finding we see most often is not a failed light. It is a building that has been testing diligently for years and writing none of it down. Undocumented testing and no testing produce identical evidence, which means all that diligence counts for nothing at the moment it is being evaluated. A one-page log with the date, the unit identifier, whether it was the monthly or annual test, pass or fail, who ran it, and what happened to the failures is enough.
What the annual test takes
The monthly test is a walk with a clipboard. The annual test is a project, and planning it as one is the difference between ninety minutes and a lost afternoon.
Find the circuit before test day, not during it
The annual test requires the units to run with no line power reaching them, which in most buildings means opening the breaker that feeds them. In a building where the panel schedule was last accurate in 1998, locating that circuit is the part that takes the longest. Sorting it out once and labeling it turns a half-day into an hour every year after.
Start the clock and leave it alone for 90 minutes
This is a duration test, so there is no shortcut version of it. Units that fail tend to fail late, at minute 60 or 70, which is exactly the information the test exists to produce. A unit checked at minute 5 and marked passing has not been tested.
Walk the egress path while it is running, not just the fixtures
Standing under each unit confirms the unit works. Walking the actual path someone would take out of the building is what tells you whether the path is lit. Dark stretches between working fixtures are a common and correctable finding, and they are invisible if you only check the fixtures.
Expect a repair list, and budget for it
Sealed lead-acid and nickel-cadmium batteries degrade quietly for years while the charging indicator stays lit. In a building that has not run a real 90 minute test recently, a meaningful share of units will not finish. That is a normal result, not a sign anything was neglected, and it is cheaper to find on test day than during an outage.
Write it down the same day
Date, unit identifier, monthly or annual, pass or fail, who ran it, and what happened to the failures. Our team provides the written record for the work we perform, and it is the document the fire code official asks for.
This is general code information, not a compliance opinion
Code sections are quoted here as published, but occupancy classification, building age, and local amendments all change how they apply to a specific property, and your fire code official is the authority on your building. What our team can tell you with confidence is what your equipment does when it is actually put under load, and what it takes to correct what fails.
Want the annual test and the repair list handled in one visit?
Our team runs the 90 minute power test on commercial and multifamily properties across Santa Clara County, walks the egress path while it is under load, replaces what fails, and leaves you the written record the fire code official asks for. Call 408-614-4451.
Common questions
How often do emergency lights and exit signs have to be tested in California?
Monthly and annually. The 2025 California Fire Code, Section 1032.10.1, requires an activation test monthly for not less than 30 seconds. Section 1032.10.2 requires battery-powered emergency lighting equipment to be tested annually by operating it on battery power for not less than 90 minutes. Both apply to illuminated exit signs as well as to emergency light units, because both are emergency lighting equipment. Records of the testing have to be kept and produced on request.
Does a self-testing fixture mean we can stop doing anything monthly?
No, and this is the most common mistake we see in buildings that spent money to avoid it. Section 1032.10.1 does allow the monthly activation test to be performed by an automated self-testing and self-diagnostic routine instead of by hand. But the same section then requires that where testing is done that way, a visual inspection of the equipment shall still be conducted monthly to identify any unit displaying a trouble indicator or that has become damaged or impaired. Self-testing changes the monthly job from pressing buttons to walking the building and reading indicators. It does not remove the monthly job.
Is the requirement 90 minutes or 60 minutes? We have seen both.
Both numbers are real and they answer different questions. For new construction, Section 1008.3.1 requires the emergency power system to provide power for not less than 90 minutes. Section 1104.5.1 provides for 60 minutes in certain existing-building situations. The annual power test in Section 1032.10.2 is written at not less than 90 minutes for battery-powered equipment. Which combination governs your building depends on its age, occupancy, and what your jurisdiction has adopted or amended, so it is worth confirming rather than assuming. Testing to 90 minutes satisfies the more demanding case.
Everything online cites NFPA 101. Why are you citing different sections?
Because California does not run on NFPA 101 as its base fire code. Buildings in Santa Clara County are inspected against the California Fire Code, Title 24 Part 9. The testing intervals happen to land in the same place, so the practical schedule you follow is the same either way. What differs is the authority. When a correction notice arrives it cites the California Fire Code section, and a printout of NFPA 101 is not a response to it. If you are building a compliance binder, cite what your inspector cites.
What if our emergency lighting runs off the building generator instead of batteries?
Then the annual test in Section 1032.10.2, which is written for battery-powered equipment operating on battery power, is not the provision doing the work for that portion of your system. Generator-backed emergency and standby power systems are inspected and tested under the emergency power system requirements, which run on a different schedule built around NFPA 110 and include load testing of the generator itself. Many buildings have both, with a generator serving the building and battery units covering specific areas, and in that case both regimes apply to their own equipment. Our team can tell you which parts of your system fall where.
How long do we have to keep the testing records?
Section 108.3 of the California Fire Code requires a record of periodic inspections, tests, servicing, and other maintenance to be maintained on the premises or another approved location for not less than 3 years, unless a different period is specified elsewhere in the code or a referenced standard, and to be made available to the fire code official. In practice that means the log lives in the building rather than in a contractor filing cabinet somewhere else.
The inspector asked for our log and we do not have one. How bad is that?
It is a correction notice rather than a catastrophe, but it is worth understanding why it lands harder than people expect. Undocumented testing and no testing produce identical evidence, so a building that has faithfully tested every month for five years and written nothing down is in the same position as one that never tested at all. Start the log now with today as the first entry rather than reconstructing dates you cannot support, get the annual test done so the first entry means something, and keep it in the building.
Can our maintenance staff do this, or does it need an electrician?
The monthly activation test is straightforward and building staff handle it in most properties. The annual 90 minute test is where people bring us in, for three reasons: it requires isolating the circuit feeding the units, it ties up someone for a full 90 minutes plus the walk-through, and it reliably produces a repair list that needs an electrician anyway. Pairing the annual test with the repairs in one visit is usually the cheaper path than running the test, discovering failures, and scheduling a second trip.
Keep reading
Commercial electrical maintenance
Scheduled maintenance for commercial and multifamily properties, including the emergency lighting testing this guide describes.
Does your building need arc flash labels?
The other compliance item that surprises property managers, and why most published guidance describes a code California has not adopted.
Can you stay open during an electrical upgrade?
How the one window where power must be off gets made short and moved somewhere it costs your operation nothing.