AWG to mm²: which metric wire, and what rounding down costs

1 mm² replaces 18 AWG. An 18 AWG conductor is 0.823 mm², so 1 mm² gives you 21.5% more copper, while 0.75 mm² gives you 8.9% less and adds 9.7% to resistance, voltage drop and heat per metre at the same current. That is with ideal copper. On data sheet limits 0.75 mm² can come out about 20% worse. Before you order either, check the terminal, because the crimp range, the insulation diameter and the seal can each rule out a size that is electrically fine.

Published advice does not settle it, even on one page. Rapid Online's chart lists 1.0 mm² as the nearest metric size to 18 AWG, its FAQ calls 0.75 mm² the nearest, and both tell you to choose 1.0 mm².¹ By area, 0.75 mm² is nearer.

Even gauges from 16 to 2 lose about a quarter when rounded down

AWG has small steps and the metric series has large ones, so the shortfall is not evenly spread. Among the even gauges, which are all that the Molex and Rapid charts list,²¹ only 20, 18 and 1/0 AWG lose under 10% of their copper. Every even gauge from 16 to 2 loses 24 to 28%, which adds 31 to 39% to resistance. The odd gauges mostly land under 10%, but check stock before you plan on one.

Area is the solid-equivalent area from the published AWG formula.³ Resistance is calculated for ideal annealed copper at 20 °C, 1/58 Ω mm²/m,⁴ with no tin and no stranding allowance. Metric sizes come from the public list of nominal sizes,⁵ and the 70 in the last row is the next size on it. That list starts at 0.5 mm². Below that the sizes you can buy come from other wire standards and catalogues. Rapid's chart gives 0.34 mm² for 22 AWG (+4.4%), 0.22 mm² for 24 AWG (+7.5%), 0.14 mm² for 26 AWG (+8.7%), 0.08 mm² for 28 AWG (-1.2%) and 0.05 mm² for 30 AWG (-1.8%).¹

What the swap does to drop and heat

At the same current, voltage drop and I²R heat per metre both move in proportion to resistance, so one percentage covers both.

As an illustration with invented round numbers, take 5 A through a 4 m loop on a 12 V supply. 18 AWG drops 419 mV, 0.75 mm² drops 460 mV and 1 mm² drops 345 mV. The round-down costs 41 mV, and whether that matters depends on how much margin the load has.

These are 20 °C figures. Copper's resistance rises by 0.393% per °C,⁴ so at 60 °C it is 1.157 times its 20 °C value: 24.24 Ω/km for 18 AWG and 26.60 Ω/km for 0.75 mm². To hold heat per metre level on the smaller wire, the current has to fall with the square root of the area ratio: a factor of 0.955 for 18 AWG to 0.75 mm², and 0.874 for 16 AWG to 1 mm². That holds heat per metre and not temperature. If the outer diameter shrinks with the conductor, the surface that sheds the heat shrinks too. As a rough illustration, current in proportion to area to the power 0.75 gives 0.933 and 0.817. Treat these as a first check on heating and not as a current rating. Bundling, ambient temperature and insulation class set the real limit, and inside a panel temperature rise is a verification of its own.

An 18 AWG wire may not be 0.823 mm²

A stranded AWG size names the equivalent solid conductor,³ and the copper you get depends on the build. Worked out from each strand diameter:

These are real products: IEWC's catalogue lists all three 18 AWG builds, and 65/30 under 12 AWG.⁶ Molex's wire chart agrees to about 0.5%, and it notes that 19-strand 18 AWG has more than 18% more material than 16-strand.²

So "18 AWG to 0.75 mm²" can mean 8.0% less copper or 22.5% less. Ask for the stranding before you take a percentage from the first table.

Metric sizes have a similar catch. An IEC 60228 size is a nominal one, set by a maximum resistance and not by measured area.⁵ One maker's H05V-K data sheet lists its 1 mm² conductor more than 10% above the ideal 17.24 Ω/km,⁷ and stranding adds resistance as well: NBS Handbook 100 quotes a 2% allowance for concentric-lay conductors.⁴ That is why the rule below starts from data sheet resistance and not from nominal area.

Watch the notation too. A US 16/30 is 16 strands of 30 AWG, 0.815 mm². A UK 16/0.2 is 16 strands of 0.2 mm, 0.503 mm².¹ The first has 62% more copper, and they look alike on a bill of materials.

The terminal decides before the wire does

Terminals are specified for ranges, and TE's general guideline says both the conductor area and the outer diameter must match the terminal drawing.⁸

Start with the barrel. TE lists one DEUTSCH size 16 pin as 20 to 16 AWG and 0.5 to 1 mm², though 16 AWG is 1.309 mm². Another is 18 to 14 AWG and 0.75 to 2 mm², though 14 AWG is 2.081 mm².⁹¹⁰ The two columns are not conversions of each other, so ask the maker which one governs before you crimp a wire that is inside one and outside the other. TE's Superseal 1.5 receptacles jump from 0.35 to 0.50 mm² straight to 0.75 to 1.50 mm²,¹¹ so on paper 0.5 mm² wire fits and 20 AWG, at 0.518 mm², does not.

A wire swap can also move the set-up. TE's data for one 3.3 mm pitch receptacle gives each wire row its own crimp height: 0.74 mm for 0.21 to 0.23 mm², 0.80 mm for 0.31 to 0.38 mm² and 0.93 mm for 0.51 to 0.53 mm². A 0.5 mm² wire is in none of those rows. Those are applicator values. The same sheet's hand tool table uses 0.75 mm for the two smaller rows and 0.90 mm for the largest.¹² Either way, a changed wire needs a fresh crimp height check and pull test.

Then the insulation. Molex says the insulation diameter of an 18 AWG wire can run from 1.78 mm to over 4.57 mm.² The seal and the insulation grip follow that diameter, not the gauge. Superseal single-wire seals are yellow for 1.7 to 2.4 mm and red for 2.5 to 3.3 mm, and the same catalogue gives its 0.75 to 1.50 mm² contact two insulation ranges: 2.40 to 3.35 mm for FLK wire and 1.7 to 2.4 mm for FLR.¹¹ Set a few real wires against those ranges:

Diameters are nominal catalogue values.⁶⁷ One colour line on the H05V-K sheet shows 2.8 mm with its conductor and resistance cells blank. That does not add up with a 0.6 mm wall, so ask the maker and measure the reel you have. Contact ranges are from TE's product sheets.⁹¹⁰

A rule for rounding down

Round up unless the terminal forbids it. Round down only if every line below is true, and write the change on the drawing.

1. Compare the maximum resistance on the two wire data sheets, and round down only if the increase is under 10%. Where a sheet gives no figure, use area against the real stranding and treat the answer as optimistic. On one maker's H05V-K sheet,⁷ 0.75 mm² is listed about 20% above ideal 16/30 copper with a 2% lay allowance. That sets a limit against an ideal figure, so it is the cautious view, but that swap can fail this line. The 10% is an editorial judgement drawn from the gap in the area figures, not a published limit.
2. Nothing else fixes the size. A protective conductor, a wire co-ordinated with a fuse or breaker, or a size named in a listing, approval or customer specification is not yours to change. Panel work has its own rules, starting with which standard applies.
3. The extra drop fits the margin. Work out current × loop length × the difference in resistance per metre, using those data sheet figures corrected to the working temperature.
4. The current is at or below the reduced figure above, using the lower factor if the outer diameter shrinks, and the wire is not already at its bundled rating.
5. The new wire is inside the terminal's area range, its insulation range and the seal's range.
6. The new wire's voltage and temperature ratings are at least those of the wire it replaces. In the table above they run from 300 V to 600 V and from 70 °C to 105 °C.⁶⁷
7. The tool is re-set for the new wire and the crimp is checked again.

On those tests, 18 AWG 16/30 to 0.75 mm² passes only if the data sheets show it, and only on a short control wire running well below its rating. 16 AWG to 1 mm² is 23.6% short by area and will not get through line 1. Rounding up has its own limit: a bigger wire can leave the barrel range and thicker insulation can leave the seal range, so lines 5 to 7 still apply.

Japanese sizes sit closer. One JIS C 3316 KIV specification lists 0.75, 1.25, 2, 3.5 and 5.5 mm².¹³ By area, 1.25 mm² is 4.5% short of 16 AWG, 2 mm² is 3.9% short of 14 AWG, 3.5 mm² is 5.8% over 12 AWG and 5.5 mm² is 4.5% over 10 AWG. Where your site can buy them, they land much closer than the IEC series by area. Check line 1 as well: on that specification's maximum resistance, 1.25 mm² is about 9% and 2 mm² about 12% above ideal 16 AWG and 14 AWG copper with a 2% lay allowance, so 2 mm² can still fail it. Check that any approval accepts the wire.

Going the other way: AWG for a metric drawing

Wire charts that list only even gauges make the reverse trip harder than the forward one.

0.5 mm² to 20 AWG and 0.75 mm² to 18 AWG are the easy ones. 1 mm² is not: 18 AWG is 17.7% short, which adds 21.5% to resistance, and 16 AWG is 30.9% over and may not fit the barrel the drawing assumed.

SWG still turns up

Rapid says SWG now survives mostly on resistance wire, fuse wire, some winding wire and older drawings.¹ The numbers are not AWG numbers. 18 SWG is 0.048 in, or 1.219 mm,¹⁴ which is 1.167 mm² and 41.8% more copper than 18 AWG. The gap changes with the number: +55.8% at 14, +58.6% at 16, +26.9% at 20, +22.0% at 22 and +19.8% at 24. If an old drawing says "18 gauge" and names no system, ask which one before you price it. BS 3737, the standard that defined SWG, has been withdrawn.¹⁴

Notes

1. Rapid Online, Wire gauge conversion chart. Supplier guide, accessed November 2026. Its chart lists even gauges only. ↩ ↩² ↩³ ↩⁴ ↩⁵

2. Molex, Quality Crimping Handbook, order no. 63800-0029, revision C, 2009, copy hosted by DigiKey. Written for Molex open barrel terminals and tooling. Its wire chart lists even gauges only. ↩ ↩² ↩³

3. Wikipedia, American wire gauge. Accessed November 2026. Used for the formula d = 0.127 mm × 92^((36 - n)/39) and for how stranded sizes are named. ↩ ↩²

4. US National Bureau of Standards, Handbook 100, Copper Wire Tables, 1966: 0.017241 Ω mm²/m at 20 °C, 0.00393 per °C at 20 °C, and a 2% lay allowance for concentric-lay conductors up to 2,000,000 circular mils. ↩ ↩² ↩³

5. Wikipedia, IEC 60228. Accessed November 2026. Used only for the public list of nominal sizes and the statement that a nominal size is set by resistance. The standard itself was not read and its tables are not reproduced here. For its resistance limits, use a licensed copy. ↩ ↩²

6. IEWC, Product Catalog, wire pages rev 3/16: PVC Type 1007, PVC Type 1015 and SRPVC Type 1061. Distributor catalogue, nominal values. ↩ ↩² ↩³

7. Klaus Faber AG, H05V-K data sheet, issue 24.03.2017, copy hosted by Conrad. ↩ ↩² ↩³ ↩⁴

8. TE Connectivity, Application Specification 114-18022, Rev K, 2011, copy hosted by Farnell. This is Rev K, the copy we could open. Check TE for a later revision. ↩

9. TE Connectivity, DEUTSCH 1060-16-0622 product sheet, copy hosted by AiChipLink. ↩ ↩²

10. TE Connectivity, DEUTSCH 1060-16-0122 product sheet, copy hosted by AiChipLink. ↩ ↩²

11. TE Connectivity, AMP Superseal 1.5 Series Connectors, catalogue revised 06-2016, copy hosted by Custom Connector Kits. ↩ ↩²

12. TE Connectivity, Application Specification 114-106094, Rev D, copy hosted by Farnell. ↩

13. Specification 010510010A-001, KIV 600 V wire to JIS C 3316, supplier specification in Japanese, hosted on Inaba's JAPPY site. Used for its list of sizes and, as percentages only, its maximum resistance. Availability in the UK or Europe was not checked. ↩

14. Wikipedia, Standard wire gauge. Accessed November 2026. Inch values as published; the areas are calculated. ↩ ↩²