The number that's right there, but you can't use it

Open the ampacity table—NEC 310.16—and a 4 AWG copper conductor stares back with three different answers. Sixty degrees Celsius gives you 70 amps. Seventy-five gives you 85. Ninety gives you 95. Same copper, same diameter, three numbers spread twenty-five amps apart. The spool in your hand is almost certainly THHN/THWN-2, stamped 90°C right on the jacket. So you reach for 95 amps, because why would the manufacturer print a rating you're not allowed to use?

And then a more experienced hand stops you: "Size it to the seventy-five column. Eighty-five."

That moment—the gap between the number printed on the wire and the number you're actually allowed to design around—is one of the most misunderstood corners of the trade. It isn't a code technicality invented to slow you down. It's a story about where wire actually fails, and the answer is almost never in the middle of the run.

Heat doesn't fail the wire—it fails the connection

A conductor's temperature rating describes its insulation, not the copper. THHN insulation can sit at 90°C continuously without breaking down. That's a real, usable property. The copper underneath would happily carry even more current; it's the plastic around it that sets the limit, because once insulation cooks, it cracks, embrittles, and eventually stops insulating.

So if the wire itself can take 90°C, why can't you load it to the 90°C ampacity?

Because the wire is not the weakest part of the circuit. The lug is. Every conductor terminates somewhere—a breaker, a panel busbar, a disconnect, a piece of equipment—and that termination is a screw, a lug, a setscrew clamping copper against metal. Those terminals have their own temperature rating, and it is usually lower than the wire's. A breaker lug rated for 75°C means: if you run enough current through this conductor to heat it past 75°C, you will cook the connection point, not the middle of the run. The lug loosens, oxidizes, develops resistance, heats further, and you've started the slow feedback loop that ends in a scorched panel.

This is the quiet logic behind NEC 110.14(C): a circuit is only as thermally strong as its terminations. You can have the most heat-tolerant wire on the market, but if it lands on a 75°C lug, the whole circuit is a 75°C circuit. The chain breaks at the link, and the link is the screw.

What 110.14(C) actually says

The rule gives you a clean default that's worth memorizing, because it covers the overwhelming majority of what you'll wire.

For circuits rated 100 amps or less, or using conductors 14 AWG through 1 AWG, you size to the 60°C column—unless the equipment terminals are listed and marked for a higher temperature, which most modern breakers and panels are (look for "75°C" or "60/75°C" stamped near the lugs). When they're marked 75°C, you get the 75°C column.

For circuits over 100 amps, or conductors larger than 1 AWG, you size to the 75°C column by default.

Notice what's missing from both: the 90°C column. In normal practice, you almost never get to terminate at the 90°C ampacity, because standard equipment terminations simply aren't rated that high. The 90°C number isn't a lie—but it's not for terminations.

So what is the 90°C column for?

Here's where the misunderstanding flips into genuine usefulness. The 90°C column exists for derating—the adjustments you make for ambient heat and for bundling conductors together in a raceway.

When you apply temperature-correction factors (a hot attic, a rooftop in the sun) or adjustment factors (more than three current-carrying conductors crammed in one conduit), you're allowed to start from the 90°C ampacity and multiply down from there. The higher starting point gives you headroom, which is the entire reason 90°C wire is worth buying.

The two-step that trips people up looks like this:

  1. Derate from the 90°C column. Take the 90°C ampacity, apply your ambient-correction and conductor-count adjustment factors, and arrive at an adjusted ampacity.
  2. Cap at the termination column. The final, after-derating ampacity is also not allowed to exceed what the termination temperature column (usually 75°C) would have given you for that wire.

You calculate from the top, but you're never allowed to land above the termination ceiling. The 90°C column is scratch paper for the derating math; the 75°C column is the wall you can't push through.

A worked example, slowly

Say you're running 6 AWG copper THHN to a 60-amp load, and the conduit happens to carry six current-carrying conductors in a 100°F (about 40°C) ambient.

  • The 90°C ampacity of 6 AWG copper is 75 amps. That's your starting number.
  • Six conductors in a raceway triggers an 80% adjustment factor. 75 × 0.80 = 60 amps.
  • The 40°C ambient applies a correction factor (for 90°C wire, 0.91). 60 × 0.91 ≈ 54.6 amps.

Now the cap. The 75°C ampacity of 6 AWG copper is 65 amps. Your derated 54.6 is below that ceiling, so the cap doesn't bind—your usable ampacity is 54.6 amps, and that 60-amp load no longer fits. You either upsize the conductor or reduce the fill.

Do the same math starting from the 75°C column instead of the 90°C column, and you'd start at 65, derate to a far smaller number, and fail much harder. That's the headroom the 90°C wire bought you—and also the reason you can't skip the cap and pretend the wire is a 90°C circuit end to end. Both the derate-from-90 and the cap-at-75 are doing real work.

Why this is easy to get backwards

The trap is that the table gives you three numbers and no narrative. Nothing on the page tells you that one column is for insulation survival, one is for the screw on the breaker, and the largest number is only ever a starting line for subtraction. Apprentices reach for the biggest number because bigger feels like more conductor for the money. Veterans who learned by rote reach for 75 every time and sometimes leave legitimate derating headroom on the table.

The mental model that fixes both: size to the connection, derate from the insulation. The termination decides your ceiling. The insulation decides your starting point. They are two different jobs done by two different columns, and the only conductor that's correctly sized is one that satisfies both at once.

There's a field-level version of this wisdom, too. The next time you torque a lug, remember that the 75°C limit assumes that connection is clean and properly tight. An undertorqued lug runs hotter than the table ever modeled, which is the same reason loose connections start fires. The column you size to and the torque you apply are the same conversation about heat, just at two ends of the same wire.

Where Voltly fits

This is exactly the kind of calculation that's easy to reason about at a desk and easy to fumble on a ladder with a 40°C attic over your head and six conductors already in the pipe. Voltly was built to keep the two-step honest: it starts the ampacity from the right column, applies your ambient and bundling factors, and then quietly enforces the 75°C termination cap so the number it hands you is one you can actually land on a lug—NEC tables, voltage drop, box fill, and conduit bends, all offline, no signal required in the basement. The point isn't to do the thinking for you; it's to make sure the column you sized to is the column the code agrees with.

If you'd rather check the cap than rediscover it on an inspection, it's at voltly.lumenlabs.works.