Checked against primary sources 2026-08-24

The 3 percent figure is a note, and knowing that is worth a mark on its own

Items are written to catch a candidate who treats a recommendation as a rule, and other items are written to catch the candidate who has overlearned that and forgets the two articles where the limit is real.

On this page

What the code actually says about it
The two articles where the limit is real
Why it gets tested anyway
The formula, both versions
Worked forward
Worked backward, which is the harder direction
The five ways this goes wrong
How to hold it under a clock
What this page cites

What the code actually says about it

One way on the tape measure, two ways in the formula. This single mistake halves more answers than any other. Voltage drop is an Informational Note, at 210.19 for branch circuits and 215.4(A)(2) for feeders, not a requirement. Exams still ask it.

For an ordinary branch circuit or feeder, nothing enforceable. The 3 percent and 5 percent figures everyone quotes sit in informational notes, one attached to the branch circuit conductor rule in Article 210 and one attached to the feeder rule in Article 215.

Those notes describe a conductor sized so the drop at the farthest outlet stays within 3 percent, with the total across feeder and branch circuit staying within 5 percent, as providing reasonable efficiency of operation. That is a description of good practice, not a requirement, and the difference is stated by the code about its own text.

NEC 90.5 sorts everything in the book into mandatory rules, permissive rules and explanatory material, and puts informational notes in the third bucket, which is explicitly not enforceable as a requirement.

So an item that asks what the code requires for voltage drop on an ordinary branch circuit has an answer, and the answer is that the code sets no percentage. A percentage sitting in the options is there to be left alone.

The two articles where the limit is real

Both are outside general wiring, both are in Chapter 5 or 6 territory, and both are written as hard numbers.

Sensitive electronic equipment

NEC 647.4(D) sets 1.5 percent as the ceiling on any branch circuit and 2.5 percent on feeder and branch circuit combined. Article 647 covers separately derived systems used to supply sensitive electronic equipment, and the tight numbers are the reason the article exists.

Fire pumps

NEC 695.7 sets two limits under two different conditions. The voltage at the controller line terminals may not drop more than 15 percent below normal under motor starting conditions, and the voltage at the load terminals may not drop more than 5 percent below the motor voltage rating with the motor running at 115 percent of its full load current rating.

Note the shape of that second one. It is measured at 115 percent of full load, not at full load, which means the calculation you run is not the one the stem first appears to be asking for. That detail is the item.

Both articles are enforceable text. Neither is a note. A candidate who has learned "voltage drop is only a recommendation" as a slogan will answer a fire pump item wrongly, which is why the slogan needs both halves.

Two ways a job binds you when the book does not

On an examination the paragraphs above are the whole answer. On a job they are not, because two things can turn guidance into obligation without changing a word of the code.

An authority having jurisdiction can amend the code it adopts, and some have written a voltage drop limit in as a local requirement. Where that has happened an inspector will hold you to it.

A specification or a contract does the same thing privately. A drawing that calls for a maximum drop is a term you agreed to. Failing it is a commercial problem rather than a code violation, which does not make it a smaller one.

The honest statement has two halves: the national code sets no general limit and says its own notes are not requirements, and what governs your particular job is the adopted code as amended where you are working, plus whatever you signed.

Why it gets tested anyway

Because it is real engineering. A conductor that satisfies every ampacity rule in the book can still deliver unusable voltage at the far end of a long run, and nothing in the ampacity calculation notices.

Motors are the usual case. Low voltage at start means more current, more heat, and a motor that can fail to come up to speed. That is a failure the derating calculation cannot see, because derating is about the conductor and this is about the load.

So the paper tests the arithmetic in a subject where the code declines to compel the outcome, and that is a perfectly coherent thing for it to do.

The formula, both versions

Voltage drop is current times the resistance of the path. The only thing the two versions disagree about is how much conductor the current sees.

Single phase: drop equals 2 times the one-way length times the current times the resistance per unit length. The 2 is the return trip.
Three phase: drop equals 1.732 times the one-way length times the current times the resistance per unit length. The square root of three replaces the 2 because the three phase currents are 120 degrees apart and their drops do not simply add.

Use the one-way length in both. The multiplier already carries the return path, and doubling the length and then multiplying by 2 as well is the commonest way this calculation goes wrong by a factor of two.

Resistance comes out of Chapter 9. Table 8 gives direct-current resistance in ohms per 1000 feet, separately for copper and aluminum and separately for solid and stranded, with coated and uncoated columns for copper. Table 9 gives alternating-current resistance and reactance for conductors in a raceway, and it is the one to use where the item mentions power factor or reactance.

Divide the per-1000-feet figure by 1000 to get ohms per foot, or keep it in thousands and divide the length by 1000. Either works. Mixing them is a factor of a thousand and you will notice.

Worked forward

Take a 240 volt single phase circuit carrying 30 amperes to a load 150 feet away on 10 AWG copper. Read the resistance for that conductor in Chapter 9, Table 8 and confirm the figure in your own book; for stranded uncoated copper it is 1.24 ohms per 1000 feet.

Convert the resistance: 1.24 divided by 1000 gives 0.00124 ohms per foot.
Apply the single phase form: 2 times 150 feet times 30 amperes times 0.00124.
That is 11.16 volts.
As a percentage, 11.16 divided by 240 is 4.65 percent.

A number that size is why the notes exist. Nothing in the code stops you installing that circuit, and no motor at the end of it will thank you.

Change nothing but the system and the arithmetic barely moves. The same current, the same distance and the same conductor on a three phase circuit gives 9.66 volts, because 1.732 has replaced the 2 and nothing else has changed. If your three phase answer comes out larger than your single phase answer on the same inputs, you have used the wrong multiplier.

Worked backward, which is the harder direction

Given a drop you are willing to accept, the question becomes what conductor delivers it. Rearrange rather than guess.

Turn the allowed percentage into volts. On the circuit above, 3 percent of 240 volts is 7.2 volts.
Solve the same formula for resistance: 7.2 divided by the product of 2, 150 feet and 30 amperes.
That gives 0.0008 ohms per foot, which is 0.8 ohms per 1000 feet.
Go to Chapter 9, Table 8 and take the first size in the right material column whose resistance is at or below that figure. Here it is 8 AWG, which clears it; 10 AWG does not.

Round toward less resistance, which means toward the larger conductor. A conductor that is one size too small fails the requirement you set yourself, and on this type of item the option list will contain that size on purpose.

One consequence worth carrying off this page: upsizing for voltage drop does not change the circuit rating, the overcurrent device or what the circuit may supply. It changes the wire and nothing else.

The five ways this goes wrong

Using one-way length with no multiplier at all. Halves the answer.
Doubling the length and then also multiplying by 2. Doubles it.
Using the single phase multiplier on a three phase circuit, or the reverse.
Answering a "what does the code require" question with a percentage.
Reading the resistance from the wrong material or the wrong construction column after correctly identifying the size.

The first three are one error wearing different clothes. Every one of them comes from not picturing the path the current takes before writing anything down.

How to hold it under a clock

Do not memorize a formula with letters in it. Remember the sentence: current times the resistance of the path, and the path is longer than the run.

From that you can rebuild the forward version, the backward version and both multipliers without needing to recall which letter stood for what. The letters are where people go wrong on a paper they are already behind on.

What this page cites

NEC 90.5 How the code sorts its own text, separating enforceable rules from permissive language from explanatory material. Informational notes fall in the third group and are not enforceable, which is the sentence the whole distinction on this page rests on. source
NEC 647.4(D) Voltage drop for sensitive electronic equipment. 1.5 percent on any branch circuit and 2.5 percent on feeder and branch circuit combined. Enforceable text, not a note. source
NEC 695.7 Voltage drop for fire pumps. 15 percent below normal at the controller line terminals under motor starting conditions, and 5 percent below the motor voltage rating at the load terminals with the motor at 115 percent of its full load current rating. source
NEC Chapter 9, Table 8 Conductor properties, including direct-current resistance in ohms per 1000 feet, by material, by construction and by coating. The source of the single value used in the worked example, which you should confirm in your own copy.
NEC Chapter 9, Table 9 Alternating-current resistance and reactance for conductors in a raceway. The table to use where an item introduces power factor or reactance rather than treating the circuit as purely resistive.