How to Check Receptacle Wiring with a Meter
Why this matters
A three-light plug-in tester answers one narrow question: is there voltage between each pair of contacts, with no current flowing. It cannot tell you how much, it cannot tell you what happens when the circuit carries a real load, and it lights "correct" on a bootleg ground, where someone jumpered the neutral to the ground screw at the device. That receptacle passes, gets signed off, and the metal case of whatever plugs into it now rides on the neutral. If that neutral ever opens upstream, the case goes toward 120 V through the load.
The meter procedure below takes three readings, twice: once unloaded and once with a known load plugged into the other half of the same duplex. It takes about four minutes a receptacle and it catches the faults the tester cannot.
Before the probes go in
This is energized diagnostic work at the receptacle face. Use a meter rated CAT III 600 V or better with intact leads and shrouded tips, prove it on a known live source before the first reading and again after the last, per NFPA 70E Article 120 (120.5 in the 2021 edition, renumbered in 2024), with work practices under 29 CFR 1910.333(b)(2). The after-check is what makes a 0.0 believable: a meter with a blown fuse or a broken lead also reads a clean zero. Wear the PPE your shop's 70E risk assessment assigns to 120 V receptacle testing, keep fingers behind the probe guards, and never hold both probes in one hand across a slot.
Any time the plan becomes "pull the receptacle and look", that is a change of work, not a continuation. Turn the breaker off, verify the box dead live-dead-live on every conductor including any second cable that may come from another circuit, and only then put a hand in the box.
Know your meter's input mode. A high-impedance digital meter (megohms) reads capacitively coupled "ghost" voltage on a conductor that is connected to nothing, often 20 to 80 V and wandering. A low-impedance mode (often labeled LoZ, a few kilohms) drains that ghost to near zero, but at 120 V it also draws tens of milliamps through whatever it is connected across. Hot to ground on a GFCI-protected receptacle in LoZ will trip the GFCI, because a 5 mA personnel GFCI sees that current leaving on the ground path. Use high-impedance mode by default and switch to LoZ only to settle a suspected ghost on a receptacle you know is not GFCI-protected.
What each of the three readings actually sees
The narrow slot is the hot (brass screw), the wide slot is the neutral (silver screw), and the round or D-shaped hole is the equipment ground. Read slot to slot at the face.
- Hot to neutral is the supply voltage minus the drop on BOTH the hot and the neutral conductors, because load current flows through both.
- Hot to ground is the supply voltage minus the drop on the hot ONLY, because a healthy equipment grounding conductor carries no current, so it sits at the panel's ground bar potential all the way out.
- Neutral to ground is the drop on the neutral only. Unloaded it sits near zero. Under load it rises by exactly the neutral's voltage drop, which makes it the one reading that tells you current is flowing in the neutral and the ground is a separate conductor.
So on a correctly wired receptacle, hot-to-ground equals hot-to-neutral plus neutral-to-ground, give or take a tenth. A receptacle that breaks that relationship, or whose neutral-to-ground refuses to move under load, is lying to the tester.
Why load the circuit at all. With no current flowing there is no drop anywhere, so a bootleg ground and a real ground produce identical numbers: 121.4, 121.4, 0.0. The only physical difference between them is that a real ground is a separate wire carrying no current. You can only see that difference when the neutral is carrying current and the ground is not.
The procedure
Read the supply first, unloaded. Hot to neutral, hot to ground, neutral to ground, in that order, on the half of the duplex you will NOT plug the load into. Write all three down to one decimal. If hot-to-neutral is well below the 114 V floor ANSI C84.1 Range A sets at the service point, look upstream before judging the receptacle.
Plug a known resistive load into the other half. A 1,500 W portable heater on high is the common choice: purely resistive, about 12 to 12.5 A, nothing electronic to confuse the readings. Clamp its cord or the circuit's hot at the panel so you know the actual current rather than assuming it. Both halves of a duplex share the same terminals unless the tab is broken, so the other half sees exactly the drop the load creates. Check first that the tab is intact: on a split receptacle the halves may be on different circuits.
Read the same three again with the load running. Let the heater settle 10 to 15 seconds, then check that hot-to-ground minus hot-to-neutral equals neutral-to-ground.
Interpret against the table below, then unplug the load. If the readings fail, the next move is a de-energized look in the box, under the isolation rule above, not a second tester.
Prove the meter again on the known live source. Every 0.0 you wrote down depends on it.
The readings, condition by condition
SAME RECEPTACLE, 65 FT OF 12 AWG FROM THE PANEL
SUPPLY HELD AT 121.4 V, LOAD 12.0 A RESISTIVE
CONDITION LOAD NAR-WIDE NAR-GND WIDE-GND
correct off 121.4 121.4 0.0
correct on 118.4 119.9 1.5
reversed off 121.4 0.0 121.4
reversed on 118.4 1.5 119.9
open ground off 121.4 ghost ghost
open ground on 118.4 ghost ghost
bootleg ground off 121.4 121.4 0.0
bootleg ground on 118.4 118.4 0.0
NAR = narrow slot, WIDE = wide slot, GND = ground hole
ghost = hi-Z meter wanders tens of volts,
LoZ reads near 0 (not on GFCI circuits)
Where the 1.5 V comes from: NEC Chapter 9, Table 8 gives 1.93 ohms per 1,000 ft for 12 AWG uncoated solid copper at 75 C. Sixty-five feet is 0.0650 x 1.93 = 0.125 ohm per conductor, and 12.0 A across 0.125 ohm is 1.5 V. The load current crosses two conductors, so hot-to-neutral falls 3.0 V, from 121.4 to 118.4. Hot-to-ground falls only the hot's 1.5 V, to 119.9. Check: 119.9 minus 118.4 is 1.5, which is the neutral-to-ground reading. A real panel also sags a little under the added load; that lowers the first two readings equally and leaves neutral-to-ground alone, so the neutral-to-ground rise stays the clean signal.
Correct. Neutral-to-ground rises by roughly the hot-to-ground drop, because hot and neutral are the same size and length.
Reversed polarity. The narrow slot now carries the neutral, so the "hot slot to ground" reading is the one near zero, and it is the one that rises under load. The tester catches this one too.
Open ground. Both readings to the ground hole are meaningless numbers on a high-impedance meter, and both go near zero in LoZ. The tester catches this too. What the meter adds is a way to tell a true open from a high-resistance ground (a loose wire nut on the bare pigtail): a high-resistance ground reads full voltage hot-to-ground in high-impedance mode and sags in LoZ once that resistance is high enough to choke the LoZ current; a few ohms of bad contact will not show here.
Bootleg ground. Unloaded it is indistinguishable from correct. Under load, hot-to-ground falls exactly as far as hot-to-neutral (118.4 and 118.4) and neutral-to-ground stays at 0.0 while 12 A flows in the neutral. On 65 ft of 12 AWG a separate ground must show about 1.5 V. A neutral-to-ground that stays at 0.0 or 0.1 V under a 12 A load is a neutral and a ground joined at or very near this box.
"Or very near" matters: neutral-to-ground measures the neutral drop back to wherever neutral and ground are joined, so at a receptacle 3 ft from the panel even a correct one reads under 0.1 V under load. The test only discriminates on runs long enough to produce a readable drop, which is why the run length goes on the ticket.
Worked example: the home office that passed the tester
A customer's home office circuit fed six receptacles. The previous owner's inspector had plugged a three-light tester into each and recorded "correct" on all six. The customer called because a new laser printer kept faulting, and wanted the circuit checked before replacing it.
The tech worked the procedure on receptacle 4, about 50 ft of 12 AWG from the panel by the route. Unloaded: 121.0, 121.0, 0.0. Heater into the lower half, clamp at the panel read 12.1 A. Loaded: hot-to-neutral 118.7, hot-to-ground 118.7, neutral-to-ground 0.0.
Expected on 50 ft of 12 AWG: 0.050 x 1.93 = 0.0965 ohm per conductor, times 12.1 A is 1.17 V, call it 1.2. So hot-to-neutral should fall about 2.3 V (121.0 to 118.7, which it did) and hot-to-ground only 1.2 V, to about 119.8, with neutral-to-ground reading 1.2. Instead hot-to-ground fell the full 2.3 V and neutral-to-ground did not move. The 2.3 V drop on hot-to-neutral says the circuit is fine as a circuit; the identical drop on hot-to-ground says the ground hole is on the neutral.
The tech turned the breaker off, verified the box dead live-dead-live, and pulled the receptacle. The box held the incoming and outgoing two-wire cables, neither with an equipment grounding conductor, and a short jumper ran from the silver screw to the green screw. Receptacles 1 to 3 measured correctly under load, each with a bare conductor in the box; 5 and 6 were downstream of 4 on the same ungrounded cable and carried the same jumper. The upstream half of the circuit had been rewired at some point and the downstream half had not.
The tech could not tie the printer faults to the jumpers and said so, rather than promising the repair would fix them.
The jumpers came out. With them gone, receptacles 4 to 6 read honestly: full voltage hot-to-neutral, ghost voltage to the ground hole, which is what an ungrounded receptacle should show. The customer chose a GFCI at receptacle 4 protecting 5 and 6, with "No Equipment Ground" and "GFCI Protected" labels, per NEC 406.4(D) in the 2023 edition (2020 where the section moved), as adopted; the alternative, an equipment grounding conductor run back to the panel, was quoted alongside. See related: How to Replace a Two-Prong Receptacle on an Ungrounded Circuit.
What changes the reading
- A multiwire branch circuit. The shared neutral carries the difference between the two legs' currents. With a load running on the other leg, neutral-to-ground at this receptacle can read less than the hot's drop, or near zero, on a correctly grounded receptacle. Before calling a bootleg on a 3-wire cable, turn off or unload the other half of the multiwire circuit and repeat step 3.
- A load that is not resistive. Motors and electronic power supplies shift the phase of the current, so the three readings no longer add exactly. Use the heater, not whatever happens to be plugged in.
- GFCI or AFCI protection upstream. A resistive heater is balanced and does not trip either device; the LoZ caution above is the only GFCI trap in this procedure.
Fault patterns beyond these four rows, such as hot-on-ground combinations or a tester that disagrees with itself, are worked through in Receptacle Tester Fault Pattern Decision Tree and Meter Says Grounded, Tester Says Open: Bootleg Ground Decision Tree.
How to know you are done
Six numbers, the load current and a rough run length per receptacle; loaded neutral-to-ground within a few tenths of the hot-to-ground drop; any failure carrying its de-energized finding and disposition; the meter proved live after the last reading.
References
- NFPA 70E, Standard for Electrical Safety in the Workplace, Article 120 (120.5 in the 2021 edition, renumbered in 2024), in the edition your safety program adopts
- 29 CFR 1910.333(b)(2), electrical safety-related work practices, deenergized parts
- NEC 2023 (2020 where the section moved), as adopted: Chapter 9 Table 8 and 406.4(D)
- See related: How to Replace a Two-Prong Receptacle on an Ungrounded Circuit
- See related: Receptacle Tester Fault Pattern Decision Tree