How often must emergency lighting be tested?

How often must emergency lighting be tested?

Three intervals, three genuinely different questions. The monthly test asks whether the lights come on. The annual test asks whether they stay on for as long as the design promised. The five-yearly photometric check asks something nobody used to be required to ask: whether the light they produce is still bright enough, in the places it needs to be bright.

The monthly function test is yours to do

The point of keeping it brief is worth understanding, because it is the part people get wrong. Emergency lighting batteries typically need a long recharge after a deep discharge, and while they are recharging the building is not covered. Running a full-duration test every month in the belief that more testing is better leaves you with worse protection, not better, and shortens the life of every battery in the building.

What you are checking is simple: does each luminaire illuminate, is each exit sign lit and legible, and on a central battery system, is the panel showing normal charge with no faults. Anything that does not light gets written down and dealt with.

The annual full-duration test

The detail that catches people out is the timing. For the hours after a full-duration test, the batteries are flat and the building genuinely has no emergency lighting cover until they recharge. So the test belongs at a point in the week where the building can afford that, which for most premises means the start of a quiet period rather than the evening before a busy one.

The other detail is that the test is judged at the end, not the beginning. A fitting that blazes away for twenty minutes and fades at two hours has failed a three-hour test, and it is a failure you cannot see unless somebody is there at the three-hour mark. That is the whole reason the test is certificated work rather than something to delegate.

New in 2025: the five-year photometric check

This is the genuinely new obligation, and most building owners have not heard of it yet. Until 2025 the regime tested whether the lights worked. It never tested whether they were still bright enough, and those are not the same question after a few years in a real building.

Light levels drift downwards for ordinary reasons, none of which show up in a function test:

• Lamps age. Output falls gradually over years. A fitting at the end of its life still lights up on a monthly test and still fails a measured reading.
• Buildings change. A new partition, a relocated door or a reconfigured office moves the escape route away from the lights that were designed around the old one.
• Surfaces change. Emergency lighting design assumes reflectance from walls, ceilings and floors. Redecorating a pale office in a dark colour genuinely reduces the light reaching the floor.
• Fittings get replaced piecemeal. A like-for-like swap that is not actually like for like leaves a gap nobody recorded.
• Dirt accumulates. Diffusers in industrial and kitchen environments lose output to a film that nobody thinks to clean.

The verification has to be done by a competent person with calibrated measuring equipment, which puts it outside what an in-house maintenance team can do. The result is recorded in the log book and compared against the original design figures, which means the other thing this change quietly requires is that somebody can still find the original design.

Three hours, or one?

In practice, sleeping accommodation, places of entertainment and any building using phased evacuation need three hours, because in all of them people are still inside long after the supply fails. The one-hour case is narrow and depends on a promise about how the building is actually used, not about how somebody hopes it will be used.

If you do not know which duration your system was designed to, that question is worth settling before the next annual test, because it decides what "pass" even means.

What the log book has to show

A log book with gaps is worse than an inconvenience: it is a written record that the building went unmonitored for a period, produced by you. The most common pattern we find is a tidy run of annual certificates from a contractor and no monthly entries at all, which tells an inspector precisely which half of the duty was being ignored.

The five-yearly photometric result now belongs in there too, along with the design figures it was measured against. If your emergency lighting has never had a photometric verification, that is not an emergency, but it is now a visible departure from the current standard, and it is the sort of thing better dealt with on your schedule than during an audit.

Emergency lighting is one part of a set that gets inspected together. The fire risk assessment is what decides where lighting is needed in the first place, and it is reviewed on its own cycle.