The Wire Colour Conventions That Are Not Universal
Why this matters
Colour is the fastest identification you will ever make and the only one you make without a tool, which is why it is the one that kills people. A conductor's colour was chosen by whoever pulled it, under whatever convention was normal in that trade, that country, and that decade, and it stays that colour long after the convention has moved on and long after somebody re-purposed the conductor for something else. Two of the colours you rely on most are permitted by code to mean something other than what you assume, and one of those permissions is why a switch box can hand you a white conductor at full line voltage.
What colour actually is
Colour is a claim, made by an installer, that a conductor has a function. It is not a property of the conductor and it is not evidence.
That does not make it worthless. It is an excellent way to generate a hypothesis and to decide which conductor to test first, and when you apply it yourself, it is how you communicate with the next tech. The failure is only ever in the last step: acting on the colour instead of on a measurement.
The code-fixed colours, and the much larger set that is convention
Under NFPA 70 (NEC), only a small number of identifications are actually fixed:
- Grounded conductors (what most people call the neutral) are identified white or grey, or by three continuous white or grey stripes, under Article 200.
- Equipment grounding conductors are green, green with one or more yellow stripes, or bare, under Article 250, and green is never permitted for an ungrounded conductor.
- Ungrounded conductors have no code-assigned colours in general. What Article 210 requires is that where a premises has more than one nominal voltage system, ungrounded conductors be identified by system, with the means of identification posted or available at each branch-circuit panelboard. The specific colours are the premises' choice.
- The high leg of a 4-wire delta system must be identified by an orange marking or other effective means where the grounded conductor is also present, under Article 110.
Everything else you know about wire colour is convention. Black, red and blue for one system and brown, orange and yellow for another is a widespread and genuinely useful convention, adopted premises by premises, and it is exactly as binding as the fact that the premises adopted it.
The convention families you will meet
| Domain | Typical scheme | Status | Where it bites |
|---|---|---|---|
| Premises branch circuits, North America | White or grey grounded, green or bare grounding, black/red/blue and brown/orange/yellow by system | Grounded and grounding are code-fixed, the phase colours are premises convention | Two systems in one building with no posted identification |
| Equipment built to harmonised colours | Brown line, blue neutral, green-and-yellow protective earth, per IEC 60445 | A different standard, correctly applied | Blue reads as an ungrounded conductor to a North American tech and is a neutral |
| Older installations built to superseded schemes | Varies by era and country, including red line with black neutral | Was compliant when installed | A colour that means the opposite of what it means now |
| Internal equipment harnesses | Manufacturer's own scheme | No standard at all | The print's colour callouts are the only authority, and they change between revisions |
| Low-voltage control wiring | Trade habit, widely shared, not a standard | Convention only | A spare conductor re-purposed at one end and not the other |
| Vehicle and low-voltage DC systems | A separate convention set again | Convention, domain-specific | A tech carrying premises-wiring assumptions into a DC system |
The two permissions that catch experienced people
White is permitted to be something other than a grounded conductor in specific cases. Article 200 covers when a white or grey conductor in a cable assembly may be used for other than a grounded conductor and what reidentification is required at the terminations. The trap is not current practice, it is history: older switch loops were run with the white conductor as the supply to the switch and no reidentification, because that was permitted at the time. That installation is still in the wall, and it still hands you a white conductor sitting at line voltage.
Bare and green are the same identification. A bare conductor is an equipment grounding conductor, not an unidentified one, and it is not a spare. Techs treat bare copper as neutral territory because it has no colour at all, which is precisely backwards.
The switch loop, worked all the way through
Replacing a wall switch with a device that requires a grounded conductor. The box contains one white and one black conductor plus the equipment grounding conductor, and nothing else.
The colour reading says white is the grounded conductor, black is ungrounded, so land the device's neutral on the white. Under the older switch-loop practice, white is the ungrounded supply arriving at the switch and black is the switched return going back to the fixture. Landing a device's neutral terminal on that white puts full line voltage on a terminal designed to sit at ground potential, and it will not blow a fuse or trip a breaker to tell you.
The test that settles it, done de-energized. Open the branch-circuit overcurrent device, lock and tag it, and prove dead with the meter on a known live source, then on the conductors, then back on the known source (NFPA 70E-2021, 120.5 for the proving sequence, with the de-energizing and locking or tagging requirement at 29 CFR 1910.333(b)(2) for work on or near exposed energized parts). Then identify by continuity rather than by colour: a true grounded conductor rings out to the grounded busbar at the panel, and an ungrounded conductor does not. Two minutes, no ambiguity, and it works regardless of what era the installation came from.
If you identify with the circuit energized instead, you are making a voltage measurement at exposed energized parts, which carries its own required work practices and its own PPE, and it gives you a less certain answer than the continuity test does. The de-energized identification is both safer and better evidence.
The tell you can see before you test. Two conductors and no others in a switch box, in an older installation, is itself the signal. A switch that only breaks an ungrounded conductor needs two conductors and no grounded conductor at all, which is why the device you are installing needs something the box was never built to supply. Noticing that before you strip anything is what turns this from a hazard into a scope conversation.
Why the convention feels reliable right up until it is not
A control enclosure inventoried during a service visit: 14 conductors in the control section. Nine of them followed the panel's own convention, which is about 64% of the 14. Three were spares re-purposed at some point, carrying a colour that described their original job. Two came from an imported subassembly built to harmonised colours, so blue in that bundle is a neutral while blue everywhere else in the same panel is not.
That is 5 of 14 conductors whose colour is misleading, and the distribution is the problem rather than the count. The 9 conforming conductors are the ones a tech meets first and most often, so the convention gets confirmed far more often than it gets contradicted. Worse, the re-purposed and imported conductors cluster in the sections that have been worked on, which is exactly where you are going.
A rule that holds about two thirds of the time in one panel is not a rule you can act on. It is a rule you can use to choose what to test first, which is exactly what a colour is for.
What to do with colour on the way out
When you leave, you own the identification you left behind.
Reidentify anything you re-purposed, at every termination, with something durable, and do it while you are at the termination rather than at the end of the job. Do not re-purpose a green or bare conductor for anything, ever. If you added conductors from a different scheme, label the bundle and say which scheme, because "blue is a neutral in this harness" is a sentence the next tech cannot derive. And where the premises has more than one nominal voltage system, the identification means has to actually be posted at the panelboard rather than living in the head of whoever wired it.
Checking your own identification before you act
One question, asked out loud: what did I test, as opposed to what did I read?
If the answer for a given conductor is "read," it is a hypothesis, and the only conductors you are allowed to act on as hypotheses are the ones where being wrong costs nothing. That is a very short list. It does not include anything you are about to land on a device terminal, anything you are about to cut, and anything you are about to touch with the circuit closed.
References
- NFPA 70 (National Electrical Code), Article 200 for grounded conductor identification and the cases where a white or grey conductor may serve otherwise, Article 250 for equipment grounding conductor identification, Article 210 for identification of ungrounded conductors where more than one nominal voltage system is present, and Article 110 for high-leg identification
- IEC 60445, identification of conductors and equipment terminals, which is the source of the brown, blue and green-and-yellow scheme
- 29 CFR 1910.333(b)(2), de-energizing and locking or tagging before work on or near exposed energized parts
- NFPA 70E-2021, 120.5, the test-before-touch sequence used to prove the conductors dead before identifying them by continuity
- See related: Cable + Wire Identification Reference; How to Read a Terminal Designation