Convert 12 AWG and you get 3.31 mm². Convert 4/0 and you get 107.2 mm². Neither is a cable you can buy.
This is not a rounding problem. The two systems were built on different principles and they were never meant to agree.
Two different design rules
AWG is a geometric series. Diameters step by a fixed ratio of 1.1229 between two anchor points, so the areas land wherever that progression puts them — 3.31, 5.26, 8.37, 13.3 mm².
The metric series was chosen. Someone picked round, memorable numbers with a roughly consistent ratio between them: 1.5, 2.5, 4, 6, 10, 16, 25, 35, 50, 70, 95.
A calculated progression and a chosen list have no reason to coincide, and they almost never do.
The round-up rule
When a converted area lands between two metric sizes — which is nearly always — take the smaller size that is not smaller than the conductor.
12 AWG converts to 3.31 mm². The sizes either side are 2.5 and 4. Take 4 mm². You get 21% more copper, which is margin you did not ask for but which costs little.
Rounding down is sometimes defensible, but check the size of the gap first. It varies enormously:
| Gauge | Round up | Round down |
|---|---|---|
| 2 AWG | +4% | −26% |
| 12 AWG | +21% | −24% |
| 14 AWG | +20% | −28% |
| 20 AWG | +45% | −4% |
| 22 AWG | +54% | no size below |
| 24 AWG | +144% | no size below |
| 4/0 AWG | +12% | −11% |
At 20 AWG the round-down costs only 4% and the round-up costs 45% — there, rounding down is obviously right. At 14 AWG the round-down costs 28%, which is a real reduction in what the circuit can carry.
Note also that the metric series stops at 0.5 mm². Anything finer than 20 AWG falls off the bottom and has no size below at all.
The test that decides it
Cable is not verified by measuring copper. Nominal cross-section is a label, not a measurement — you cannot put calipers on a conductor and read 4 mm².
What gets tested is DC resistance per kilometre at 20 °C. Every nominal size has a maximum it must not exceed.
This is where “12 AWG equals 4 mm²” falls apart. A 12 AWG conductor runs about 5.21 Ω/km. The limit for 4 mm² is 4.61 Ω/km. It fails by 13%.
So the substitution runs one direction only:
- Replacing an AWG design with metric cable? Round up. You get equal or better performance and you meet the limit for what you ordered.
- Someone offering an AWG conductor against a metric spec? It will usually fail the resistance test. This is the one that costs money, because it surfaces at a factory acceptance test rather than at the quote.
The two exceptions
Almost every gauge fails the limit of the size it rounds up to. Two do not:
| Gauge | Rounds up to | Jump | Actual R | Limit | Result |
|---|---|---|---|---|---|
| 2 AWG | 35 mm² | +4% | 0.51 Ω/km | 0.524 | passes |
| 2/0 AWG | 70 mm² | +4% | 0.26 Ω/km | 0.268 | passes |
The reason is visible in the numbers. Where the step up is only 4%, the conductor is close enough to nominal to stay inside the limit. Everywhere else the jump is 11% or more and it fails.
Two kcmil sizes behave the same way — 350 kcmil against 185 mm², and 750 kcmil against 400 mm².
If you need an AWG conductor that genuinely satisfies a metric specification, these are the four sizes where it happens.