Emergency light testing is two jobs wearing one name. One is short, frequent and boring: prove the fitting comes on when the power goes. The other is long, annual and the one that finds the failures: prove it stays on for its rated duration, which in the United States is generally ninety minutes. Almost every problem a small contractor or facilities manager has with emergency lighting comes from treating the two as one job, or from doing both and being unable to show it afterwards. This page separates them and says what each one leaves behind.
The monthly test proves the fitting switches, and almost nothing else
A monthly functional test is a key turn or a breaker drop and a look: does the lamp light, is the lens clear, is the charge indicator healthy. It takes a minute or two per fitting and it catches the failures that are already obvious, which is worth doing because obvious failures are also the ones a visitor notices. What it cannot tell you is whether the battery behind the light has any capacity left, because a cell with fifteen minutes in it lights just as convincingly for thirty seconds as a cell with two hours.
The annual test is where the failures actually show up
The annual test runs the fittings on battery for their full rated duration, and the Minnesota Department of Health's guidance on battery-operated emergency lights, citing NFPA 101 section 7.9.3, states that a minimum of 90 minutes of illumination is required. Ninety minutes is wall-clock time somebody is on site for, which is why it is the visit that slips and why it is the visit that finds things. On the worked example in the free emergency lighting test log on this site, that single annual visit is a larger share of the year's hours than several monthly rounds put together.
An inspection and a test are not the same word to everyone
Emergency lighting inspection, emergency light inspection and emergency light testing are used interchangeably by most buyers and distinguished carefully by some standards, where inspection means the visual check of the installation and testing means operating it. If you are writing a schedule, define which you mean in it. If you are buying software, ignore the distinction: what you need recorded is the same either way, which is the device, the date, the result and the person.
The part nobody buys software for is the part they get asked about
Almost every contractor can do the test. Far fewer can produce, in May, the log for the November round on a site they visit four times a year. The record is the deliverable, and it has to be device-level: a note saying the site passed is worth very little when the question is about one fitting in a stairwell. The free test log on this site sizes the work; keeping it against the devices is what the paid plan is for.
The self-testing alternative, and what it does not remove
Fittings with self-test circuitry run their own monthly function test and their own annual duration test, and show the result on an indicator; a central system may log them. The code the guidance cited below follows recognises those as alternatives to a person turning a key. What they do not remove is the record: somebody still reads each fitting's indicator, records the result against that fitting, and acts on the failures, and a central log still has to be kept and produced. In the free test log on this site the share of fittings that self-test is an input for that reason. It cuts the minutes on the walk; it does not cut the visits or the rows.
Questions people ask about emergency light testing
How often does emergency lighting have to be tested?
The frequency your building has to meet is set by the code your authority having jurisdiction enforces, not by a website. The pattern most US buildings follow is a short functional test monthly and a full-duration test annually, and the Minnesota Department of Health guidance cited below sets out those options as NFPA 101 section 7.9.3 gives them. Ask your AHJ or your insurer which edition applies to you before you write your schedule.
Why ninety minutes?
Because it is the duration the code is written around: the guidance cited below states that a minimum of 90 minutes of illumination is required. It exists so that the lighting outlasts an evacuation with a margin, and it is long enough that degraded batteries fail during it, which is exactly why the shorter test cannot replace it.
Do self-testing fittings mean I do not have to do anything?
No. A self-testing fitting runs its own test and shows you the result; someone still has to read the result, record it against that fitting, and do something about the ones that failed. It removes the key turn, not the record. In the free test log on this site, the share of fittings that self-test is an input for exactly that reason: it changes the minutes, not the visits.