Emergency lighting testing under NFPA 101: the monthly test proves nothing about the annual duty
NFPA 101 section 7.9.3 requires a 30-second functional test every 30 days, spaced 3 to 5 weeks apart, and an annual test of not less than 90 minutes on battery. Pressing the test button proves the lamp lights. Only the 90-minute discharge proves the battery carries the escape route.
By Hovermarks team
Quick answer. NFPA 101 section 7.9.3 sets two tests on two cadences. A functional test every 30 days for not less than 30 seconds, with the interval running not less than 3 weeks and not more than 5 weeks between tests. And an annual test of not less than 90 minutes for battery-powered equipment, run on battery alone. Self-testing or computer-based systems may be used in place of the manual test, with automatic 30-day testing and an annual 1.5-hour test. Written records of all inspections and tests must be kept for review.
Walk into most commercial buildings and the emergency lighting register looks healthy. Twelve monthly entries a year, ticks all the way down, someone's initials against each. Then ask for the annual duration test and the conversation changes, because the two are not versions of the same thing and a register of one cannot evidence the other.
The monthly test proves the lamp works
Pressing the test button, or turning the key, cuts mains power for a moment and confirms the unit illuminates on battery. That is a real check and it catches real faults: dead lamps, disconnected batteries, units that were switched off during a refit and never switched back.
What it cannot tell you is whether the battery holds. A cell degraded to four minutes of capacity passes a 30-second test perfectly. It illuminates, the tester ticks the box, and the register records a pass that is entirely truthful and completely uninformative about the duty the code actually imposes.
The annual test proves the escape route stays usable
The 90-minute test is a different question: can this unit light the egress path, on battery alone, for the period the code requires. That takes running the unit to the end of the duration under real discharge, which is why the annual test is a duration test rather than a functional one.
Two things worth understanding about why the number is what it is:
- A unit can stay lit while its output decays. Illumination falls as the battery drains, and a luminaire technically still on can be well below a usable level for anyone finding an exit in smoke. The code is about a usable escape route for the whole period, not a lamp that has not yet gone out.
- Batteries fail gradually and invisibly. There is no external sign that a four-year-old battery now delivers 20 minutes. The only thing that surfaces it is a full discharge, once a year, on purpose, at a moment of your choosing rather than during an actual evacuation.
The 3-to-5-week window, which is not the same as monthly
This is the detail that trips teams up, and it is stated explicitly in the code. The functional test happens every 30 days, with not less than 3 weeks and not more than 5 weeks between tests.
"Monthly" in ordinary usage means "once in each calendar month", and that is not what the code says. Consider two teams:
- Tests on 1 January and 5 February: 35 days apart. Inside the window, compliant.
- Tests on 1 January and 28 February: 58 days apart. Both are "monthly" in the everyday sense, both appear as one entry per calendar month on the register, and the gap is outside the code's maximum.
The second pattern is common wherever the test drifts to whenever someone has time, and it produces a register that looks complete on a calendar view. Any system tracking this properly measures the interval between tests rather than counting entries per month.
What the record has to hold
Owners are required to keep written records of the inspections and tests for review. In practice a defensible record per unit holds:
- Which test this was, functional or annual duration, since conflating them is the failure this article is about
- The date, and for the functional test the interval since the last one
- The duration achieved on the annual test
- Whether the unit remained adequately illuminated throughout, which is a different question from whether it stayed on
- Any unit replaced or battery changed, and when
- The person who carried out the test
Where Hovermarks fits
Hovermarks ships two separate templates for this, on every paid plan:
- Monthly Emergency Lighting Inspection, the functional check
- Annual 90-Minute Emergency Lighting Duration Test (NFPA 101 7.9.3)
They are deliberately two templates rather than one with extra questions, because they are different tests on different intervals producing different evidence, and folding them together is exactly how a register ends up unable to answer the annual question.
Two design decisions worth naming, because they follow from the argument above:
- The elapsed duration is recorded as a number, not a pass or fail. A unit that failed at 62 minutes and one that failed at 4 minutes are both failures and are not the same problem. One is a battery approaching end of life; the other is a unit that would have been dark before anyone reached the stairwell. A bare pass/fail throws that distinction away.
- A separate question asks whether the egress route stayed adequately illuminated throughout, not merely whether the unit stayed on, because those are different observations and only one of them is what the code is protecting.
Each luminaire and exit sign is its own QR-tagged asset, so the register is per unit rather than per building, and the monthly and annual cadences schedule independently against each one. Hovermarks produces and schedules the record; the duty to keep it, and the judgement about whether a unit passed, stay with the owner and whoever ran the test.
The fire safety solution shows the workflow, and the free NFPA ITM schedule builder will put emergency lighting alongside the rest of a building's ITM calendar.