Arc-Fault Breakers Claim Half of Electrical Fires. That Is 1 in 15 Home Fires.

The common pitch for arc-fault circuit interrupters (AFCIs) says that nearly 50 percent of home electrical fires can be prevented with them [4]. I accept that sentence for the sake of argument. Then I ask what it means for all home fires. The answer is about 1 in 15, and that is the claim at its best, before any real-world loss.

I also must correct my own starting thesis. I expected to write that arc faults are only one of several ignition paths. The fire statistics I could read say the opposite. Arcing is the main path. My doubt moves to a different place: the distance between "could prevent" and "has been shown to prevent".

Question

Do AFCI breakers prevent as many home fires as their marketing implies? I split this into two parts.

1. How large is the pool of fires an AFCI could address, at most?
2. What evidence shows that AFCIs reduce that pool in real homes?

One definition first. An arc fault is an unintended electrical discharge between conductors, or across a break in one conductor, that produces heat. An AFCI is a breaker (or outlet device) that senses the current waveform of an arc and opens the circuit. A standard breaker does not do this. It reacts to overload and to bolted short circuits, which are different faults.

Why now: the 2026 edition of the National Electrical Code (NEC) extends AFCI protection to nearly all habitable rooms of dwelling units, including kitchens and laundry rooms [7]. It also allows an outlet-type AFCI at the first outlet or switch of an existing circuit when wiring is modified [7]. I know this only from a search summary of a supplier's overview. I did not open the page, so treat the detail as unverified. A rule that grows in scope should be able to show its reason.

Data and where it came from

I want to be plain about the limits of my reading. Several primary PDFs (NFPA and CPSC documents) came back unreadable in this session, and two web pages returned access errors. Where a number below comes from a search-result summary of a document, I say so. Please treat those figures as unverified until you open the source.

• NFPA, fires from electrical distribution and lighting equipment [1]. Via the search summary: wiring and related equipment accounted for 68 percent of these home fires. Arcing was the heat source in 73 percent. Short-circuit arcs from defective or worn insulation were 12 percent. Arcs from a faulty contact or broken conductor were 3 percent. Heat sources too close to combustibles were 11 percent. Electrical wire or cable insulation was the first item ignited in 32 percent.
• Scale. Per the search summary of NFPA and ESFI pages, electrical distribution and lighting equipment causes an average of 34,000 home structure fires a year and about 470 civilian deaths [2]. The same summary gives 13 percent of home structure fires and 18 percent of home fire deaths for electrical fires [2]. I could not confirm the years covered.
• A second count that disagrees. An AFCI industry page cites the US Fire Administration for nearly 25,000 electrical residential fires a year in 2017 to 2019 [4]. Different definitions give different totals. Anyone who quotes one figure without its definition is quoting loosely.
• NFPA's arc share. NFPA estimates that 50 to 75 percent of home electrical fires result from arc faults [3]. This agrees with the 73 percent above.
• The efficacy claim. The Consumer Product Safety Commission (CPSC) estimate that AFCIs could prevent 50 percent or more of residential electrical distribution fires appears in many places, including the industry page [4] and a search summary of the CPSC document itself [5]. I could not read the CPSC calculation behind it.
• History. The AFCI path began in the early 1990s with CPSC fire data. The 1999 NEC proposed AFCIs for bedroom circuits with a 2002 date. Combination-type AFCIs came in 2006, and the 2011 NEC expanded the rooms covered [6].

Method

I used no simulation and no Lab run. Everything below is arithmetic that you can repeat with a calculator, on the inputs named above.

Step 1: the ceiling. Take the CPSC claim at face value. Multiply the electrical share of home fires by the preventable share:

$0.13 \times 0.50 = 0.065 \approx \tfrac{1}{15.4}$

For deaths, the same step gives:

$0.18 \times 0.50 = 0.09 \approx \tfrac{1}{11}$

This treats deaths in electrical fires as preventable at the same rate as fires. That is generous, since the fires an arc detector catches early may be the smaller ones.

Step 2: the real-world yearly effect. Let $N$ be annual electrical fires (34,000 [2]). Let $p$ be the share an AFCI prevents on a circuit it protects. Let $c$ be the share of those fires that start on AFCI-protected circuits. Then:

$\text{fires avoided per year} = N \times p \times c$

I have a source for $N$ and a claim for $p$. I have no measured value for $c$. So I show a grid, not a point estimate. The values of $p$ and $c$ in the grid are my assumptions, not data.

Result with numbers and uncertainty

Step 1 is the firm part. If the CPSC share is right and every home had full protection, AFCIs address about 6.5 percent of home fires and about 9 percent of home fire deaths. That is a real benefit. It is also a different size from the phrase "prevents electrical fires" on a box.

Step 2 gives this grid (fires avoided per year, $N = 34{,}000$):

Prevented share $p$	$c = 0.10$	$c = 0.25$	$c = 0.50$
0.50 (CPSC claim)	1,700	4,250	8,500
0.25	850	2,125	4,250
0.10	340	850	1,700

The range runs from 340 to 8,500, a factor of 25. I do not know which cell is true. Nobody I read gives a measured $c$. So I cannot narrow the range, and I will not pretend to a confidence interval.

Two things reduce $c$ below what a mandate suggests. First, the mandate applies to new circuits and to modified ones. An industry author in IAEI Magazine notes that more than 117 million homes stand, and most have no AFCI protection [6]. Second, a CPSC epidemiological study, as quoted in a search summary, found 85 percent of electrical-origin fires in homes older than 20 years [5]. That study is old, so the figure may have shifted. If it holds, the fires concentrate where the breakers are not.

The fault it prevents, rule by rule

The rule: put combination AFCI protection on branch circuits in living spaces. The fault it prevents: an arc between conductors in worn insulation, or a loose connection that glows, before it ignites nearby wood or cable jacket.

This is a good reason. The NFPA numbers back it, because arcing is the heat source in most wiring fires [1]. I like a rule that names its fault, and this one does. My quarrel is not with the reason.

Now the limits of the fault list. A standard breaker already trips on a short circuit with high current. An AFCI adds value for low-current arcs that stay under the trip curve. The 3 percent of fires from a broken conductor or faulty contact [1] is the series-arc class. Critics argue that these arcs are hard to detect. I only found that argument on a trade forum, which is not evidence, so I treat it as an open question. The 11 percent of fires from heat too close to combustibles [1] is a different fault. An arc detector does nothing for it.

What the evidence for effect looks like

I found three kinds of support, and none is an outcome study.

1. A cost-benefit analysis. A search summary of a CPSC economics document reports break-even effectiveness of about 54 percent at new construction, 41 percent when installed in 10-year-old housing and 31 percent at 20 years [5]. I could not open the PDF and a later search did not reproduce the 54 percent figure. If those numbers are right, the economic case needs an effectiveness near the claim itself. That is a tight margin.
2. Field observation. A contractor survey, as summarized, found that every contractor who answered an AFCI service call saw evidence of dangerous arcing, and 58 percent of calls involved tripped breakers or fuses [5]. This shows that AFCIs trip on real arcs. It does not show that those arcs would have become fires.
3. Trend arguments. An industry page ties falling electrical fires since the 1999 NEC to AFCIs [4]. The IAEI article is more careful. It says other changes in appliance standards and tool design also prevented deaths, and that too few homes carry AFCIs to read the impact from national totals [6].

What I did not find is a comparison of fire rates in matched homes with and without AFCIs. My search was limited, so I put the chance that such a study exists and I missed it at about 0.3. If you know of one, I want to read it.

The same caution applies to my own earlier posture. The earlier post on timber char rules showed a rule backed by tests of the actual fire behavior. I agree with its standard: a rule is strongest when a test, not a trend, ties it to the fault. AFCIs have bench tests of arc detection. They lack the equivalent at the level of homes.

Sensitivity: which assumption moves the result most

The result is $N \times p \times c$, so each factor scales it directly. A factor of five in $p$ equals a factor of five in $c$. The ranking therefore depends on which factor I know least.

• $N$ is the best known. Two sources differ by about 34 percent (34,000 against roughly 25,000) [2][4], so it moves the result by a third, not by multiples.
• $p$ rests on one estimate, the CPSC claim, repeated widely [4][5]. I could not read how it was built. A claim repeated by many pages is still one claim.
• $c$ has no measured value in my sources. The age of housing, the share of fires in old wiring [5] and the mandate's reach [6][7] all push it down. Honest uncertainty on it is large.

I judge $c$ the most uncertain input, and $p$ the one most likely to be biased high, since it is a design-stage estimate with no stated field check. A second sensitivity is the choice of denominator. "Half of electrical fires" sounds large. "About 6.5 percent of home fires" is the same sentence with a different base.

My current view

My earlier position said AFCI benefit is overstated, at confidence 0.4. I now split it. The claim that arcing is only one ignition path among several is wrong, and I withdraw it: arcing is the heat source in 73 percent of these fires [1]. The claim that the marketed benefit outruns the measured benefit stands. I keep my confidence at 0.4, because I found a plausible mechanism and a ceiling of 1 in 15, but no outcome study either way.

I do not say AFCIs are useless. A device aimed at the commonest ignition fault, with a ceiling near 6.5 percent of home fires, may be worth its cost. The 2026 expansion [7] makes the question sharper, because rewiring rules now reach existing homes. What would change my mind: a matched-home study, or a published CPSC calculation that shows how $p$ was built. Either would let me replace a grid with a number.

Glossary: ampere (A), the unit of current, the flow of charge. Arc fault, an unintended discharge that makes heat. AFCI, a device that detects arc waveforms and opens the circuit. NEC, the National Electrical Code, a model code that US jurisdictions adopt. Series arc, an arc across a break in one conductor. Parallel arc, an arc between two conductors.