Electrical

Safe Isolation Procedure

Safe isolation is the single most important thing you do all day: identify, isolate, secure, prove dead with a GS38 two-pole tester against a known live source, then re-prove the tester — miss a step and you're working on something you only assume is dead.

Updated 20 Aug 20268 min read

More electricians are hurt working on circuits they believed were dead than on ones they knew were live. Safe isolation is the procedure that turns an assumption into a proven fact, and it is deliberately built as a fixed sequence so that no single failure — a mislabelled board, a borrowed neon screwdriver, a flat tester battery — can let you touch a live conductor.

This guide walks through the full procedure as set out in the HSE guidance (GS38 for test equipment and HSG85 for working safely), the Electricity at Work Regulations 1989, and the IET On-Site Guide: identify the circuit, isolate it, secure and lock it off, prove dead, and prove the tester before and after.

Key takeaways

The sequence is fixed: identify, isolate, secure/lock off, prove dead, re-prove the tester — never skip a step.
Use an approved GS38 two-pole voltage indicator, not a neon screwdriver or a multimeter set to volts.
Prove your tester works on a known live source (or a proving unit) immediately before and after proving dead.
Lock off with a padlock and unique key, and fit a caution notice so no one re-energises the circuit.
Prove dead between all conductors: line-neutral, line-earth and neutral-earth (all combinations).

Why the procedure exists

The Electricity at Work Regulations 1989 require that conductors are dead before work starts unless it is unreasonable for them to be, and that precautions are taken to prevent them being made live again. Safe isolation is how you discharge that duty in practice — it is not a formality, it is the legal and practical basis on which you are allowed to put your hands on the installation.

The reason it is a rigid sequence rather than a checklist you dip into is that each step guards against a different failure. Identifying guards against isolating the wrong circuit; locking off guards against someone switching it back on; proving the tester before and after guards against a dead tester giving you a false all-clear. Drop one and the others no longer protect you.

The one that kills

The step people skip is re-proving the tester after proving dead. A tester that failed mid-test will show a dead circuit as dead — and a live one as dead too. Prove it again on a known source.

Step 1: identify

Identify the exact circuit or item you are going to work on, and confirm it at the distribution board. On a well-labelled board this is quick; on a poor one, do not trust the schedule alone — verify by switching the circuit and confirming the load drops out, or trace it, so you isolate the conductor you actually mean to.

This is also the moment to confirm the point of isolation you will use: a circuit breaker you can lock, a fused connection unit, a main switch, or a dedicated isolator. Isolating at a point you cannot secure is not isolation.

Steps 2 and 3: isolate, then secure and lock off

Isolate at the chosen point — switch off the breaker, withdraw the fuse, or open the isolator. Then secure it: fit a lock-off device and a padlock with a unique key that stays in your pocket, and attach a caution notice stating that the circuit is being worked on and must not be switched on. Where a device cannot be physically locked, use an approved lock-off kit designed for it.

The point of the lock and key is single-point control: nobody but you can restore the supply, and you cannot restore it by accident. On a multi-person job every person working on the circuit fits their own lock, so the supply cannot be restored until the last person removes theirs.

Key stays with you

One lock, one key, one pocket. A padlock with the key left in it isn't a lock-off — it's a suggestion.

Step 4: prove dead (with the tester proven live first)

Before you go near the isolated conductors, prove your GS38-compliant two-pole voltage indicator works: test it on a known live source — a proving unit or a point you know is live — and confirm it indicates voltage. Only a tester you have just seen work is fit to prove something dead.

Now prove the circuit dead at the point of work. Test between every combination of conductors — line to neutral, line to earth, and neutral to earth — so a single crossed or borrowed connection cannot hide a live conductor. The indicator must show no voltage on every combination. A GS38 two-pole tester is the correct instrument: it has fused, shrouded probes with minimal exposed metal, and it does not rely on a battery to register mains voltage the way a multimeter or a neon does.

Step 5: re-prove the tester

Immediately after proving the circuit dead, return to the known live source or proving unit and confirm the tester still indicates voltage. This is the step that closes the loop: it proves the tester was working throughout, and that the "dead" reading you just took was a genuine absence of voltage rather than a failed instrument.

If the tester fails to indicate on the known source at this point, your dead reading is worthless — you cannot trust it, and you must re-prove with a working tester before touching anything. Only once the tester has been proven live, proven the circuit dead, and been re-proven live is the isolation complete and the work safe to begin.

Prove — dead — prove

Say it as three beats: prove the tester live, prove the circuit dead, prove the tester live again. Same known source both times.

Frequently asked questions

Why can't I use a neon screwdriver to prove dead?

A neon screwdriver is not GS38-compliant and is not a reliable voltage indicator — it can show dead when a circuit is live, and relies on current flowing through you. Use an approved two-pole voltage indicator with fused, shrouded probes.

Why prove the tester before and after?

Proving it before confirms the tester works so you can trust a dead reading. Proving it after confirms it was still working when you took that reading — if it failed mid-test it would show a live circuit as dead. Both are needed; the second is the one most often skipped.

What voltage combinations do I test when proving dead?

Test every combination at the point of work: line to neutral, line to earth, and neutral to earth. Checking all three means a crossed connection or a shared conductor can't leave a live part undetected.

Do I need to lock off, or is switching off enough?

You must secure the isolation. The Electricity at Work Regulations require precautions against the circuit being made live again, so lock off with a padlock and unique key and fit a caution notice. Simply switching off leaves the circuit one careless hand away from being re-energised.

From guidance to action

EICR certificate appRecord inspection and test results on site after safe isolation.