Reporting & Records
10.2 million incidents. Here's what they say about your department.
The 2026 Fire Service Index, ESO's annual analysis of fire department data, covered 10,237,656 incidents from 3,805 agencies during 2025, the final full year of NFIRS reporting before NERIS. The headlines: fires are 3 percent of what fire departments do, EMS is two-thirds, response-time averages hide a long tail, and the biggest documentation gaps sit exactly where money and firefighter health are decided. Below: the six findings worth acting on, and how to run your department's numbers against them.
Key takeaways
- EMS and rescue made up 66 percent of call volume; fires just 3 percent, and the EMS number is probably an undercount, which matters because grants and budgets follow documented volume.
- Median fire response beat the NFPA 1710 turnout benchmark (71 seconds vs. 80), but 90th-percentile response ran past 17 minutes, more than double the median. Percentiles, not averages, tell the truth.
- 72 percent of outdoor fires had no cause recorded, and complete on-scene decon documentation appeared in about 2 percent of exposure incidents, the two gaps with the most compliance and health exposure.
- NFIRS captured only 19 EV and e-bike fire incidents nationally, scattered across 10 incident codes, while 44,840 generic vehicle-fire records likely buried the rest. NERIS adds a dedicated classification.
- The report's own best advice: don't consume the national numbers, benchmark your department against them, station by station.
What is the Fire Service Index?
An annual data report from ESO, drawn from departments contributing to its data collaborative. The 2026 edition (10,237,656 incidents from 3,805 agencies during calendar 2025) is the largest dataset the report has compiled. Two honest caveats it carries: the data represents agencies on one vendor's platform rather than a national census, and reporting practices vary between departments. Read it as directional benchmarking, not gospel. That's also exactly how it's most useful.
The timing gives it extra weight: 2025 was the last complete year of NFIRS data before NERIS became the national standard, making this the closing chapter of a 50-year data era.
Finding 1: Fire departments barely fight fires
EMS and rescue: 66 percent of incidents. Good intent calls, service calls, and false alarms: 27 percent. Actual fires: 3 percent. No firefighter is surprised, but boards and town councils often are, and the number carries a warning: the Index notes the EMS figure is probably low, because some departments stopped generating separate EMS records during the reporting transition. Undercounted volume is invisible workload, and invisible workload loses budget arguments. The fix is procedural: every EMS run gets its own record, no exceptions.
Finding 2: Structure fires are a residential problem
79 percent of structure fire responses were residential, and 77 percent of those were single- and two-family dwellings. 61 percent of structure fire responses were working fires, up four points from the prior edition. Meanwhile sprinkler documentation barely exists: present in 7 percent of records, absent in 82, undetermined in 11, numbers that haven't budged year over year, which weakens every sprinkler-ordinance argument the fire service tries to make.
Finding 3: Wildland fire season is now a calendar year
Outdoor fires peaked in March, but they occurred in every month, even December logged roughly 3,700 wildland incidents. The documentation gap is worse than the seasonality: 72 percent of outdoor fires had no cause recorded. Cause data drives prevention programs and mitigation grants; a blank field funds nothing.
Finding 4: Medians look fine. Percentiles don't.
Median fire turnout was 71 seconds, beating NFPA 1710's 80-second benchmark. Median travel was 4:25 against the 4-minute reference, and total median response came in just under 7 minutes. Then the tail: 90th-percentile fire response ran past 17 minutes, more than double the median. Turnout on confirmed working fires (76 seconds) actually ran slower than nonworking calls (69), crews gear up for the real thing. The takeaway for chiefs: report percentiles by station, not department averages. Averages flatter; the 90th percentile is where residents actually live.
Finding 5: Decon documentation is improving, from terrible
77 percent of incidents with documented fire or smoke exposure included at least one decontamination procedure, up from 75. But across roughly 29,000 exposure incidents and 135,000 individual exposures, all four recommended on-scene decon steps were documented together just 651 times, about 2 percent. The improvement came from a software behavior, not a culture change: validation rules that require decon fields before an exposure incident can close. Given what exposure records mean for cancer presumption claims decades later, this is the cheapest insurance in the report, the Index pegs implementation at roughly $150 per station.
Finding 6: EV fires are invisible in the old data
NFIRS identified 19 electric vehicle and e-bike fire incidents nationally, scattered across 10 different incident type codes, while 44,840 generic vehicle-fire records almost certainly buried the real number. NFIRS simply had no code for the incident type defining this decade's hazard conversation. NERIS fixes it with a standardized Electrical System Storage classification; until your department is fully transitioned, a local custom incident type keeps EV and lithium-battery fires countable instead of invisible.
What should your department do with this?
Five moves, all record-keeping, all cheap: audit that every EMS run generates its own record; enforce cause-field completion on outdoor fires; turn on validation rules for decon documentation; create custom EV/lithium-battery incident types; and run the comparison station by station, department-wide averages hide exactly the variance you need to see. Every one of these is a configuration decision in decent records software. This is what RedAlert's custom reporting is for.
Common questions
Where does this fire department data come from?
From the 2026 Fire Service Index, ESO's annual report analyzing 10,237,656 incidents submitted by 3,805 agencies on its platform during calendar 2025. It's the largest dataset in the report's history, and still a sample of one vendor's customers, not a national census.
Does EMS really make up two-thirds of fire department calls?
Yes: 66 percent in this dataset, and the true figure is probably higher, since some departments stopped generating separate EMS records during the reporting transition. If your department's EMS share looks low, audit the records before assuming you're the exception.
Why are property loss numbers changing under NERIS?
NERIS replaces field-estimated loss values with insurance-sourced data. Expect your loss figures to shift for reasons that have nothing to do with your fireground performance, worth briefing your governing body before the numbers move.
How do we track EV fires before we're fully on NERIS?
Create local custom incident types in your RMS for EV, e-bike, and lithium-battery incidents. Otherwise they disappear into generic vehicle-fire codes, which is how a national hazard produced 19 documented incidents in the old system.
What's the most useful thing to do with national benchmarks?
Run your own numbers against them, by station, not department-wide. The value isn't knowing the national median turnout is 71 seconds; it's knowing which of your stations beats it and which one is 40 seconds behind.
National data is interesting. Your data is actionable.
See how departments benchmark station-level response times, close documentation gaps, and build the report the board asked for.