Guide
AWG to mm²: There Is No Exact Match
Conversion tables print "12 AWG = 4 mm²" as though the two systems line up. They do not. 12 AWG is 3.31 mm², which is 17% less copper than a 4 mm² cable — and if you substitute one for the other you have quietly changed the voltage drop. This explains why the systems cannot line up, where the gaps are worst, and what to size on instead.
AWG is a count, not a size
Metric conductors are named after their area: 4 mm² has a cross-section of four square millimetres. The name is the measurement.
AWG is not. It counts how many times a wire was pulled through progressively smaller dies to reach that thickness. More passes means thinner wire, which is why the numbers run backwards — 14 AWG is thinner than 4 AWG — and why the very thick sizes ran out of numbers and became 1/0, 2/0, 3/0, 4/0.
Because each pass reduces the wire by a fixed ratio, the areas form a geometric series rather than round numbers. Every three gauges roughly halves the area: 4 AWG is 21.15 mm² and 10 AWG is 5.26 mm², six steps and almost exactly a quarter. That regularity is elegant, and it is exactly why no gauge ever lands on a metric size. Two systems built on different principles cannot agree except by coincidence.
How far apart they actually are
Run the numbers and the gaps are larger than the tables suggest.
14 AWG is 2.08 mm² against a 2.5 mm² cable — 17% less. 12 AWG is 3.31 against 4 mm², again 17% less. 8 AWG is 8.37 against 10 mm². 6 AWG is 13.3 against 16 mm². Through the whole common range, the American conductor is the smaller one by 12 to 17%.
The two closest matches in the range are 2 AWG at 33.63 mm² against 35, and 2/0 at 67.43 against 70 — both under 4% apart, and both still under.
So a table that maps 12 AWG to 4 mm² is not converting. It is naming the next size up.
The direction of the error flips
This is the part that catches people, and it is why a rule of thumb is dangerous here.
From 14 AWG up to 4 AWG, every gauge is smaller than the nearest metric size. Then it inverts. 1 AWG is 42.41 mm², which is 21% larger than a 35 mm² cable. 1/0 is 53.48 against 50, 7% larger. 4/0 is 107.22 against 95, 13% larger.
So "round to the nearest metric size" gives you a bigger conductor than specified in the small sizes and a smaller one in the large sizes, or the reverse, depending entirely where in the range you are. There is no consistent safety margin in either direction. Anyone applying a single mental adjustment across the range will be wrong at one end of it.
Size on the requirement, not on the name
The mistake is treating this as a translation problem. It is a selection problem.
Work out what the circuit actually needs — current-carrying capacity for the installation method and ambient temperature, and voltage drop over the real run length. That gives you a required area. Then pick the nearest available size in whichever system your supplier stocks, rounding up.
That approach is indifferent to which system the drawing was written in, and it never produces the failure this guide is about: a conductor 17% smaller than the one specified, chosen because a table said the two names were equivalent.
If you must record a cross-system equivalence, write both the gauge and the area, and note which one governs.
Common questions
What is 12 AWG in mm²?
3.31 mm². Tables commonly print 4 mm², which is the next size up and 17% more copper. For sizing, work from the area the circuit needs rather than from the equivalence.
Cable sizes and resistance →Is 10 AWG the same as 6 mm²?
No. 10 AWG is 5.26 mm², about 12% less than 6 mm². That is one of the closer pairs in the common range, and it is still a real difference in a long run.
Can I just round to the nearest metric size?
Not safely, because the direction changes. Below 4 AWG every gauge is smaller than its nearest metric size; from 1 AWG upwards several are larger. 1 AWG is 21% above 35 mm². A single mental adjustment will be wrong at one end of the range.
Why do AWG numbers get smaller as the wire gets thicker?
Because the number counts drawing passes, not size. Each pass through a smaller die makes the wire thinner, so more passes means a higher number and less copper. Past the thickest single-digit size the count runs out, which is where 1/0 through 4/0 come from.
Does stranding change the calculation?
Slightly. The strands spiral, so each is a little longer than the cable and resistance rises by roughly 2% over the solid-conductor figure. It matters on long runs that are already near the voltage drop limit.
Voltage drop calculator →