Skip to content

Equipment Grounding Conductor Size Chart

The whole of NEC Table 250.122 — all 19 rows to 6000 A, copper and aluminum — with the 250.122(B) math for circuits whose conductors you already enlarged.

NEC Table 250.122, 250.122(A), 250.122(B)

Overcurrent Device & Material

100 A · Cu#8 EGC

These are the ratings Table 250.122 actually prints. A device rating that is not on the list — 50 A, 125 A, 175 A — reads off the next row up, never the one below.

Grounding conductor material

The aluminum column covers aluminum and copper-clad aluminum alike. Note that the two materials are not one size apart at every row — the gap widens as the table climbs.

Read this table by the device, not the wire

Table 250.122 is indexed by the rating of the overcurrent device protecting the circuit. It is not indexed by the size of the circuit conductors, and looking up your wire gauge in it is the single most common way to get an answer that is wrong in both directions. A #6 feeder on a 60 A breaker takes a #10 EGC; the same #6 on a 200 A service does not.

Minimum Equipment Grounding Conductor

Minimum EGC#8Copper, 100 A deviceNEC Table 250.122

Both materials at this row

Copper#8
Aluminum / copper-clad aluminum#6

250.122(A) — an EGC never has to beat the circuit conductors

Whatever the table returns, the equipment grounding conductor is not required to be larger than the ungrounded conductors supplying the equipment. That ceiling holds over the proportional rule below as well, so a scaled requirement that exceeds the circuit conductor stops at the circuit conductor.

Based on NEC 2023. Sizes here are minimums for the grounding path — verify against the edition your jurisdiction has adopted before you pull wire.

Verified against NEC 2023 · Table 250.122, 250.122(A), 250.122(B)

Did you upsize the conductors? Then the EGC moves too — NEC 250.122(B)

The table minimum assumes the ungrounded conductors are the size the device rating implies. Go up a size or two for voltage drop on a long run and that assumption breaks: the larger conductors have lower impedance, so a ground fault at the far end drives more current back through a return path that has not changed. The grounding conductor is that return path, and it is what has to hold long enough for the breaker to trip. 250.122(B) requires it to grow in the same proportion as the circular mil area of the ungrounded conductors — not by the same number of AWG steps, which is where this goes wrong by hand.

52,620 cmil

211,600 cmil

Sizing the run itself is a different job — the voltage drop calculator tells you how far up you have to go, and this tells you what that costs you in grounding conductor.

The 2020 exception

The 2020 edition added a permission for a qualified person to size the grounding conductor under 250.4(A)(5) or 250.4(B)(4) instead — an engineered showing that the fault path is low-impedance enough to clear the device. That is judgement, not arithmetic, so this panel does not model it. Absent that engineering, the proportion is the rule.

Upsized EGC#1up from the #8 table minimum

Derivation

Minimum EGC for a 100 A deviceNEC Table 250.122#8
Minimum ungrounded conductor #3NEC Chapter 9, Table 852,620 cmil
Installed ungrounded conductor #4/0NEC Chapter 9, Table 8211,600 cmil
Proportional increaseNEC 250.122(B)× 4.021
#8 EGC (16,510 cmil) scaled by that ratioNEC 250.122(B)66,391 cmil
Smallest conductor meeting that areaNEC 250.122(B)#1 (83,690 cmil)

NEC 2023 Table 250.122 — minimum size equipment grounding conductors

All nineteen published rows, 15 A through 6000 A. The left column is the rating or setting of the overcurrent device ahead of the circuit — the row you are on right now is highlighted. Print this page for a field copy; the controls above drop out and the table does not.

NEC Table 250.122, minimum size equipment grounding conductors for grounding raceway and equipment, by overcurrent device rating
Device ratingamperes, not exceedingCopperAluminumor copper-clad aluminum
15 A#14#12
20 A#12#10
60 A#10#8
100 A#8#6
200 A#6#4
300 A#4#2
400 A#3#1
500 A#2#1/0
600 A#1#2/0
800 A#1/0#3/0
1000 A#2/0#4/0
1200 A#3/0250 kcmil
1600 A#4/0350 kcmil
2000 A250 kcmil400 kcmil
2500 A350 kcmil600 kcmil
3000 A400 kcmil600 kcmil
4000 A500 kcmil750 kcmil
5000 A700 kcmil1250 kcmil
6000 A800 kcmil1250 kcmil

Watch the bottom two aluminum rows. 5000 A and 6000 A read 1250 kcmil, not 1200. The 1200 kcmil figure stood for more than fifty years and was revised upward in the 2020 edition because it no longer cleared a fault at those ratings — yet reference tables published today still print it while claiming to be current. If you are copying this row off another site, check which one it gives you.

Where this table does not reach. It covers grounding conductors run with the circuit — raceways, cable trays, equipment enclosures. It is not the grounding electrode conductor, which is Table 250.66 and sized off the service conductors instead, and it is not the bonding jumper on the supply side. Different tables, different inputs, routinely confused.

How this works

Table 250.122 sizes the equipment grounding conductor, the wire that carries fault current back to the source if something goes wrong, and it reads off the breaker or fuse protecting the circuit, not the wire gauge you're pulling. That trips people up constantly: the same 6 AWG feeder wants a 10 AWG ground behind a 60 amp device and a 6 AWG ground behind a 200 amp device, because the table is really rating how much fault current the device will let through before it opens.

The rows aren't evenly spaced, they jump from 20 to 60 to 100 to 200 amps, so a device rating with no row of its own, 125 amps or 175 amps, takes the next row up. Rounding down to the nearer row undersizes the ground for the fault current the device actually allows.

Upsizing the circuit conductors for voltage drop upsizes the ground too, proportionally, under NEC 250.122(B). The rule scales by circular mil area rather than counting gauge steps, and the two methods give different answers: going from 3 AWG to 4/0 is a 4.02x increase in area, which scales an 8 AWG ground up to 1 AWG, not the 2 AWG a gauge-step guess would land on. The one hard ceiling on all of it is NEC 250.122(A): the ground never has to be larger than the circuit conductors themselves.

Worked example

A 100A circuit whose ungrounded conductors get upsized from #3 AWG (the minimum for the load) to 4/0 AWG for voltage drop.

Minimum EGC for a 100A deviceNEC Table 250.122#8 AWG copper
Minimum conductor (#3 AWG) areaNEC Chapter 9, Table 852,620 cmil
Installed conductor (4/0 AWG) areaNEC Chapter 9, Table 8211,600 cmil
Proportional increaseNEC 250.122(B)× 4.021
#8 AWG (16,510 cmil) scaled by that ratioNEC 250.122(B)66,391 cmil
Smallest conductor meeting that area (#2 is 31 cmil short)NEC 250.122(B)#1 AWG

The grounding conductor scales from #8 AWG to #1 AWG, not the #2 AWG a two-gauge-step guess would land on.

Frequently Asked Questions

What size equipment grounding conductor does a 100 amp circuit need?

Number 8 copper, or number 6 aluminum. Table 250.122 gives one pair of sizes per device rating and those are the 100 ampere row. The rows are not evenly spaced — the table jumps from 20 to 60 to 100 to 200 amperes — so a rating with no printed row of its own, such as 125 or 175 amperes, takes the next row above it. Reading downward to the nearer row leaves the grounding path undersized for the fault current the device is willing to let through.

Do you look up Table 250.122 by the wire size or by the breaker size?

By the breaker or fuse, always. The left column is the rating or setting of the overcurrent device ahead of the circuit, and the conductor gauge you happen to be pulling does not enter into it. This is the mistake that shows up most in the field: the same 6 AWG feeder wants a 10 AWG grounding conductor behind a 60 ampere device and a 6 AWG one behind a 200 ampere device, because what the table is really rating is how much fault current the device will permit before it opens.

If I upsize conductors for voltage drop, does the ground wire have to grow too?

Yes, and in proportion — NEC 250.122(B). Enlarging the ungrounded conductors lowers their impedance, so a fault at the far end of the run pushes more current back down a return path that has not changed size. The rule scales the grounding conductor by the ratio of circular mil areas, not by counting gauge steps, and the two give different answers. Going from 3 AWG to 4/0 is a factor of 4.02 on area: an 8 AWG grounding conductor scales to 66,391 circular mils, which 2 AWG misses by 31, so the answer is 1 AWG. The 2020 edition added an exception letting a qualified person size the conductor under 250.4(A)(5) or 250.4(B)(4) instead, on an engineered showing about the fault path.

Can the equipment grounding conductor end up larger than the circuit conductors?

It is never required to be. NEC 250.122(A) caps it at the size of the ungrounded conductors supplying the equipment, and that ceiling survives the proportional increase — where the arithmetic asks for more conductor than the circuit itself carries, the circuit conductor is the answer. Nothing stops you installing something larger if you want to; the code sets a floor and this ceiling, not an exact size.

Why do some charts show 1200 kcmil aluminum at 5000 and 6000 amperes?

Because they are reprinting a pre-2020 table. That figure was unchanged for over fifty years and the 2020 edition revised it upward: a 1200 kcmil aluminum or copper-clad aluminum conductor is no longer accepted as sufficient at those two ratings, and the requirement is now 1250 kcmil. Reference sites still publish the old number under a banner claiming compliance with the 2020, 2023 and 2026 codes. If you are copying the bottom of this table off somewhere else, check which value you are being handed.

Is this the same table as the one for the grounding electrode conductor?

No — different table, different input, and they are routinely mixed up. Table 250.122 sizes the conductor that runs with the circuit to ground equipment enclosures and raceways, and it reads off the overcurrent device. The grounding electrode conductor, which bonds the system to the earth at the service, comes from Table 250.66 and reads off the size of the service-entrance conductors. Sizing one from the other gets you a number that is defensible-looking and wrong.