Fire Response Targets Look Like a Cliff. The Data Shows a Slope.

Plain-language summary. The 4-minute and 8-minute marks in the main US fire standard are targets that a department must meet 90 percent of the time. They are not points where people start to die. The studies I could read show fire damage rising gradually with response time. I could open only abstracts and one official statistics release, so I label every number below as abstract or release text.

The question

Does fire outcome fall off a cliff at a specific response time? Or does it slope?

Fire standards read like cliffs. NFPA 1710 sets a first-engine travel time of 4 minutes and a full-force travel time of 8 minutes, at least 90 percent of the time [6]. The memo that describes these marks calls them travel times. The clock a resident experiences starts earlier, at the call. So the "8 minutes" on a council slide and the resident's clock are different clocks.

My earlier ambulance post found that the 8-minute ambulance target traces to cardiac arrest research, where minutes without CPR or a defibrillator do cost survival. I held that fire has no such biological cliff. This post tests that. I extend the earlier post, and I do not claim fire is the same case.

Data and where it came from

I looked for studies that relate response time to a fire outcome. I read three, with limits:

• Buffington and Ezekoye (2019). US NFIRS data for fires in 1 and 2 family homes, 2002 to 2017. I read the abstract only [1].
• Manes and Rush, University of Liverpool. English Home Office fire statistics, 2010/11 to 2016/17. I read the abstract only [2].
• GOV.UK fire statistics, year ending March 2026. I read the release text [3].

I also checked the government toolkit page. It says the toolkit uses formulae that link response time to loss of life and property. It gives no coefficients [4]. The 2012 update and a 2010 research report were PDFs I could not read. I will not describe their formulae.

The NIST crew-size experiments are a different kind of evidence. They timed 22 fireground tasks in a purpose-built house. Four-person crews finished 30 percent faster than two-person crews [5]. That is task speed in a training facility, not lives saved. The source says so itself [5].

Method

I did not run code. Everything below is quoted from an abstract or release, or is reasoning shown in the text.

My grade rule, written before I grade:

Grade	Design	Effect	Certainty
Strong	Several independent studies, outcome measured directly	Effect size with interval, same direction	Full text read
Mixed	One or two studies, or proxy outcome	Direction clear, size unclear	Abstract or secondhand
None	Rule has no stated study	None	None

Result: the evidence has a slope

US homes. Across response times of 3 to 13 minutes, every averaged severity measure rises [1]. The authors tie this mainly to a higher chance of very severe fires at longer times. The strongest links are with the share of fires doing "extreme" damage to at least one story and with the share of property value lost [1]. A monotone rise across ten minutes is a slope. The abstract gives no coefficients, and it does not report whether any one minute shows a jump. That is what I need and do not have.

English buildings. The Liverpool abstract reports an upward trend in casualty rates for "other buildings" as response time rises. It also says the more severe damage classes increase in the 6 to 8 minute band for those buildings [2]. This is the nearest thing to a threshold in my sources. It is a band in one building category, not a line, and it sits on the English average of roughly 8 to 9 minutes for those buildings [2]. A band near the average has the most fires in it. Counts there will look large for that reason alone. I cannot tell from the abstract whether the authors adjusted for it.

Official English statistics. In the year ending March 2026, the average response to a primary dwelling fire was 8 minutes 15 seconds. Fires with victims or rescues averaged 7 minutes 57 seconds [3]. The release states: "There is no straightforward relationship between response times and the outcome of a fire." [3] It also warns that fires with victims are only about 9 to 10 percent of dwelling fires, so their averages swing [3]. And 78 percent of dwelling fires were attended within 10 minutes [3].

One more point from that release. Crews reach fires with victims faster, because call handlers send help sooner when someone is trapped. That reverses cause and effect in a naive comparison. Fast arrival marks a worse fire. This is why raw averages hide the slope.

Threshold table (partial)

Outcome	Threshold in source	Design	Source type	Unit
Severity of home fires	None stated; rise over 3 to 13 min [1]	Observational, US national data	Abstract	Per fire, by minute
Damage in other buildings	6 to 8 min band flagged [2]	Observational, England	Abstract	Per fire, by band
Dwelling fire outcome	"No straightforward relationship" [3]	Official statistics	Release text	Averages by year
Task speed by crew size	Not a time threshold [5]	Field experiment	News summary	Per crew

The units differ in every row. Per fire by minute, per fire by band, yearly averages and per crew cannot be compared without conversion. I leave them unconverted. My full threshold table, with a unit audit, is still queued.

What failed, what helped

What failed. The standard presents a clean pair of numbers. I found no source that derives 4 and 8 from fire evidence. One critic argues the standard rests on the wrong measures, and he cites flashover at about six minutes. His source for that is not given in the article [7]. I also found no derivation in the other sources. That absence is from my reading, not proof that no derivation exists.

What helped. Studies that treat time as a continuous variable gave the clearest picture [1]. Also, the English release states plainly that the relationship is not straightforward [3]. That sentence is worth more than many slide decks.

Evidence grades

• The 4- and 8-minute marks as survival limits. Grade: none. I found no study that shows a limit at either point.
• Faster response lowers fire damage, as a gradual effect. Grade: mixed. Direction agrees across [1] and [2]. Size is unclear. I read abstracts only.
• Larger first crews speed fireground work. Grade: mixed. The experiment is controlled but uses one house. It measures tasks, not deaths [5].

Sensitivity: which assumption moves the result most

Four assumptions could change my view, in order of weight.

1. Reverse causation. Fires with victims get faster dispatch [3]. If a study does not handle this, the slope could be flat or even negative in some cuts. This is the largest risk.
2. Which clock. Travel time, dispatch-to-arrival and call-to-arrival start at different moments. A threshold on one clock could vanish on another.
3. Fire size at the call. A fire that has already spread when someone calls will damage more at any response time. Delay in discovery is not delay in response. The Liverpool work studies both [2], which is why I want its full text.
4. Binning. If data are cut into minute bands, a smooth slope can look like steps. The 6 to 8 minute flag may be this effect [2].

My current view

I think the cliff is a product of how standards are written. A 90 percent compliance rule needs a number, and a number invites a cliff reading. The evidence I could read points to a slope with weak detail on its shape. I hold this at about 0.7. It is lower than I would like, because I read abstracts. I also distrust new measures by habit, and I read failure better than success, so I discount my own enthusiasm.

Two results would change my mind. One is a full-text study showing a clear break at a single minute in a cohort with controlled call time. The other is a derivation of 4 and 8 from fire growth data in the NFPA committee records. If either exists, I will say so and grade again.