We are going to disappoint you slightly with the headline answer, and then explain why the disappointing answer is the useful one.
Why batteries fail invisibly
An emergency light is the only fitting in your building that is designed never to be on. That creates a peculiar problem: there is no day-to-day signal that anything is wrong.
A failing fluorescent tube flickers. A failing motor gets noisy. A failing emergency lighting battery does absolutely nothing differently, right up until the moment it is needed and cannot deliver.
And it does not fail cleanly. It loses capacity while retaining the ability to work perfectly for short periods. A fitting with a badly degraded battery will:
- Illuminate instantly when you flick the test switch
- Look completely normal for the first several minutes
- Collapse well short of its rated duration
Which means the monthly check — the one somebody actually does — is structurally incapable of finding the most common failure. That is not a criticism of the monthly check. It is what the full-duration test exists for.
We are not going to publish a single number of years here. Battery service life varies materially with chemistry, fitting design, ambient temperature and how often the system has been discharged. A figure lifted from a website and applied to your building is exactly the kind of confident-sounding number this site exists to avoid. Use the manufacturer's stated life for the battery actually in your fitting, and treat it as a planning input rather than a guarantee.
The cohort problem
This is the part that catches building owners financially, and it is entirely predictable.
A building gets fitted out. Every emergency light goes in over a few weeks, with batteries from the same batch, commissioned at the same time, living in the same conditions.
Some years later they reach end of life together.
So the failure pattern is not a trickle of one or two fittings a year. It is a single inspection that turns up dozens of failures at once, an unbudgeted five-figure bill, and — if it happens in the wrong month — a Form 12A that cannot be issued until it is remediated.
Owners experience this as a sudden crisis. It is not sudden. It was scheduled the day the building was commissioned; nobody wrote the date down.
What shortens battery life
Worth knowing, because several of these are within your control.
Heat. The big one. Batteries in ceiling spaces, near heat-producing plant, in unventilated voids or in west-facing rooms age faster than the same battery in a cool corridor. If one part of your building consistently fails first, look up.
Frequent deep discharge. Every full discharge uses some of the battery's life. This is a real argument against running discharge tests more often than your compliance schedule requires, and a strong argument against running two close together.
Incomplete recharge. A battery repeatedly returned to service before it is fully recharged does not reach full capacity and degrades faster. This is why the recharge window after a discharge test matters and why back-to-back testing is a bad idea.
Age on the shelf. A battery degrades from manufacture, not from installation. Fittings that sat in a store for a year before install started life partly used.
Power quality. Sustained supply issues affect the charging circuit as well as the battery.
Why the discharge test is the only real check
Given all of the above, the full-duration discharge test is not a compliance formality. It is the only diagnostic you have.
The test runs each fitting on battery for the duration your building is designed to provide, and — critically — checks at the end whether it is still producing adequate light. Not whether the lamp is still lit. Whether the output is still adequate.
Two things follow.
The duration must be the right one. It is not automatically 90 minutes. Building Code clause F6 sets it by risk group, and many ordinary New Zealand buildings are 30-minute systems. Testing a 30-minute system for 90 minutes needlessly ages every battery in the building. See how long emergency lighting must stay on.
Shortening the test destroys its value. Battery failures cluster in the last third of the discharge. A test stopped early skips precisely the window where the information is.
Recording battery age per fitting
This is the practical fix for the cohort problem, and it costs nothing but discipline.
Against every fitting on your register, record:
- Date the fitting was commissioned
- Date the battery was last replaced — and by whom
- Battery type and manufacturer's stated service life
- Discharge test results, with the duration achieved, not just pass or fail
- Any failure and what was done
The field that changes your life is the second one. With battery replacement dates recorded per fitting, you can sort the register by age and see, at a glance, what is coming and when. Without it, you find out at the inspection.
This is also why we record replacements against the fitting rather than as a line on an invoice. An invoice tells you what you spent. The register tells you what you own.
Planning staged replacement
Once you can see the ages, the strategy writes itself.
- Sort the register by battery age. Find your cohorts.
- Identify the fittings approaching stated service life in the next 12–24 months.
- Replace them in planned tranches — by floor, by zone, by budget year — rather than waiting for a failure round.
- Prioritise the hot locations. Ceiling spaces, plant areas, anywhere that has failed early before.
- Stagger deliberately. If you replace a whole cohort at once you have simply recreated the cohort. Splitting a large replacement across two budget years breaks the cycle permanently.
- Re-test and record every replacement against the fitting.
The financial argument is simple. Staged replacement is a predictable line in a maintenance budget. Reactive replacement is an unplanned bill arriving at the least convenient moment, often alongside a BWoF that cannot be issued — see what happens if you miss a BWoF inspection.
If you have inherited a building and nobody can tell you when the emergency lighting batteries were last replaced, assume you are due a cohort failure and get a full-duration test done before your next BWoF rather than after. Finding it in your own time is a maintenance job. Finding it at the inspection is a compliance problem.
Related: why buildings fail emergency lighting on their BWoF and how often emergency lighting needs testing.