Fire Department Response Time in 8 U.S. Cities: What 2025 Incident Records Show

Fire department response time in eight cities, ranked by the 90th percentile

Ranked on the slow end of the distribution, Cincinnati, Raleigh and New York City reach 90% of fires within about six and a half minutes on their own clocks. Los Angeles and San Jose sit far behind, partly because their clocks and categories differ.

The 90th percentile describes the slow calls, not the typical one, and it is the form both NFPA 1710 and the Austin Fire Department's 8-minute goal use: a time met for 90% of incidents (City of Santa Monica summary of NFPA 1710; standard page at NFPA 1710).

Read the ranking in two groups. San Francisco (7:01) and Louisville (7:53) time from the 911 call, and Detroit (7:16) and San Jose (14:00) from a call or event stamp (called_at, Date_Time_Of_Event). Cincinnati (4:46) and New York City (6:23) start when the incident is entered in fire dispatch, and Raleigh (6:20) starts at dispatch, so their figures leave out some or all call processing.

Where each city starts and stops the clock

Every city stops the clock at the first unit on scene, but the start ranges from the 911 call to the moment a unit is dispatched. That choice alone can move a result by half a minute or more.

First-unit response time: The interval from the city's start stamp (call received, CAD entry or dispatch) to the first fire unit recorded on scene. On-scene stamps are entered by crews on vehicle terminals, which LAFD notes depends on how promptly the button is pressed.

Call processing: Time from the 911 call to dispatch. LAFD's published monthly average was 1:05 in 2025. New York City's published averages imply that 911 call-taking before FDNY receives a structural fire call adds about 31 seconds (5:06 end-to-end against 4:35 dispatch and travel, fiscal 2025).

New York City figures here use FDNY's field definitions: the interval runs from incident creation in the StarFire dispatch system to the first unit on scene. Cincinnati's data dictionary describes its start stamp as the moment response data is submitted into CAD. San Francisco's dataset defines its start as the call received at the 911 Dispatch Center.

The slow tail: median against 90th percentile

Typical calls are close together across cities; the slowest tenth is where the gaps open up.

Medians cluster between 4:23 (Raleigh) and 5:10 (Detroit) for five of the eight cities. The gap from median to 90th percentile was 101 seconds in Cincinnati, 118 in New York City, 151 in San Francisco and 192 in Louisville, but 285 in Los Angeles and 360 in San Jose.

A wide gap means a minority of incidents wait much longer than the typical one. Possible reasons include calls at the edge of station areas or a first-due company already committed elsewhere, but the records show the size of the gap, not its cause.

Drive time against the NFPA 1710 240-second travel objective

NFPA 1710 sets its engine arrival objective on travel time only. Five cities publish an en-route stamp; one of them clears the 90% mark.

As summarized by the City of Santa Monica, NFPA 1710 calls for an engine company to arrive within a 240-second travel time for 90% of incidents. NFPA now lists NFPA 1710 as combined into NFPA 1750, 2026 edition. We do not reproduce NFPA text; the comparison below uses only that published objective.

San Francisco's first-arriving units drove 240 seconds or less on 96.9% of 2025 structure fire calls, with a 90th percentile of 3:04. Louisville reached 81.1%. New York City's 59.4% is not like for like: FDNY's travel field starts at first assignment, so it includes the time crews take to turn out.

Sprinkler systems act before any unit is on scene; their inspection and testing intervals are covered on our fire sprinkler inspection page.

Share of fires reached within 8 minutes

On an 8-minute line, 6 of 8 cities reached at least 90% of 2025 fire incidents on their own clocks.

The 8-minute line comes from the Austin Fire Department, whose published goal is a first arriving unit within 8 minutes of phone pickup, 90% of the time. Austin's own file reports 69.3% of all emergency incidents (not only fires) within 8 minutes in 2024. Austin is not in our ranking because it publishes this metric only in aggregate.

At a stricter 6-minute line, used here only as a sensitivity check and not a standard, the shares fall to 96.6% in Cincinnati, 85.3% in New York City, 80.4% in San Francisco, 72.1% in Detroit, 41.9% in Los Angeles and 20.2% in San Jose.

How much response time varies inside one city

District-level 90th percentiles inside a single city can differ by more than the gap between whole cities.

New York City's five boroughs sit within 54 seconds of each other at the 90th percentile. San Francisco's ten battalions span 96 seconds and Detroit's seven council districts 75 seconds. Los Angeles shows the widest range, 527 seconds across 101 station districts.

New York City: alarm-company signals, building types and boroughs

New York City's file is large enough to split by how the fire was reported and what kind of building burned.

Phone-reported structural fires had a 2025 median of 3:50 and a 90th percentile of 5:21 (12,983 incidents). Incidents with the source code CLASS-3 (11,158 incidents) had a median of 5:12 and a 90th percentile of 6:59. FDNY's data dictionary does not define CLASS-3; the Mayor's Management Report separates fires reported through private fire alarms from 911 calls, and we read CLASS-3 as those alarm transmissions. Treat that reading as an inference.

For property managers, the practical point is that a monitored alarm and a 911 call are different routes into dispatch, and the clock for an alarm-company signal is recorded differently. Monitoring and testing requirements are covered on our fire alarm inspection page; New York code adoption is on the New York page.

By building class, Other Commercial Building Fire incidents (FDNY's classification list has no separate office class) had a 2025 90th percentile of 6:50 (3,639 incidents), against 6:40 for private dwellings. Office-specific fire safety duties are summarized on our offices page.

What changed since 2019

Five cities got slower at the 90th percentile (Louisville measured from 2021) and three got faster; two of the comparisons are weakened by changes in coding.

Los Angeles' 90th percentile for Non-EMS emergencies rose from 9:32 in 2019 to 11:13 in 2025, and LAFD's own published structure fire average rose by 30 seconds over the same years. San Jose's rose from 11:54 to 14:00. Raleigh's fell from 8:31 to 6:20, though with about 210 building fires a year the Raleigh figure moves easily.

San Francisco's change is not a clean comparison: the call type was Structure Fire in 2019 (4,791 calls) and Structure Fire / Smoke in Building in 2025 (1,902 calls), a much narrower set.

How our numbers compare with what departments publish

Where a department publishes an average for the same interval, our recomputation matches it closely, which supports the method.

FDNY reports a fiscal 2025 average of 4:35 for structural fires (dispatch and travel); our calendar 2025 mean from the incident file is 4:35 (Mayor's Management Report). LAFD publishes a monthly Non-EMS operational average; the 2025 months average 7:17, against our 7:16 (LAFD citywide metrics).

Records used and records dropped

Exclusions are small in most cities but reach a fifth of building fires in Louisville and an eighth of incidents in Los Angeles and Cincinnati.

We dropped incidents with no on-scene or start stamp, zero or negative intervals, and intervals over 60 minutes. In Louisville, 80 of 395 building fire incidents in 2025 list only support apparatus (unit type 60) and no engine, truck, quint, rescue or chief car arriving. In Los Angeles 9,147 incidents lack any on-scene stamp, and LAFD notes that its en-route and on-scene stamps depend on crews pressing a key on the vehicle terminal.

The numbers in one list

1. Cincinnati had the shortest 90th percentile of the eight cities in 2025, 4:46 from CAD entry to the first CFD apparatus on scene (1,008 structure fire incidents).^[8][9]
2. Raleigh's 90th percentile from dispatch to arrival at building fires was 6:20 in 2025 (209 incidents); Raleigh publishes no call-receipt stamp, so call processing is not included.^[13]
3. New York City's 90th percentile for structural fires was 6:23 in 2025 across 24,240 incidents, up 14 seconds from 6:09 in 2019.^[1]
4. San Francisco's 90th percentile from the 911 call to the first unit on scene was 7:01 in 2025 across 1,874 structure fire calls.^[4]
5. Detroit's 90th percentile for building fires was 7:16 in 2025 (1,542 incidents), 24 seconds faster than in 2019.^[11]
6. Louisville's 90th percentile for building fires was 7:53 in 2025 (315 incidents with a suppression unit on scene), up 68 seconds from 2021.^[12]
7. Los Angeles' 90th percentile for all Non-EMS emergency incidents was 11:13 in 2025 (66,269 incidents); LAFD's raw file does not separate structure fires.^[5]
8. San Jose's 90th percentile for structure fires was 14:00 in 2025 (431 incidents), with a median of 8:00; San Jose stamps times to the minute.^[7]
9. 6 of 8 cities had the first unit on scene within 8 minutes for at least 90% of 2025 fire incidents; Los Angeles (71.6%) and San Jose (52.7%) did not.^{[8][13][1][4][11][12][5][7]}
10. San Francisco's first-arriving units drove 240 seconds or less on 96.9% of 2025 structure fire calls, above the 90% NFPA 1710 objective; Los Angeles reached 36.3% for Non-EMS incidents.^[4][5][15]
11. Within New York City in 2025, the 90th percentile ran from 5:45 in Brooklyn to 6:39 in the Bronx and Queens.^[1]
12. Within Los Angeles in 2025, the 90th percentile ranged from 8:54 to 17:41 across 101 station districts with 30 or more incidents.^[5]
13. In New York City, CLASS-3 structural fire incidents had a 2025 median of 5:12 against 3:50 for phone-reported fires.^[1]
14. Our New York City mean of 4:35 for calendar 2025 equals the 4:35 fiscal 2025 average in the Mayor's Management Report, and our Los Angeles mean of 7:16 is within one second of LAFD's published 7:17.^[1][3][5][6]

Calculations

Download the calculations (CSV)

Summary

• Across eight cities, the 2025 90th percentile from the start of each city's clock to the first unit on scene ran from 4:46 in Cincinnati to 14:00 in San Jose.
• Cities whose clock starts at a call or event stamp (San Francisco, Detroit, Louisville, San Jose) posted 90th percentiles of 7:01, 7:16, 7:53 and 14:00.
• New York City's 90th percentile for structural fires was 6:23 in 2025, and its mean of 4:35 matches the 4:35 average FDNY reports for fiscal 2025.
• Drive time is where the NFPA 1710 objective applies: San Francisco (96.9%) met the 90% mark; Louisville (81.1%), New York City (59.4%), San Jose (43.3%) and Los Angeles (36.3%) did not.
• The slow tail differs more than the typical call: the gap between median and 90th percentile was 101 seconds in Cincinnati and 360 seconds in San Jose.
• Since 2019 the 90th percentile rose by 101 seconds in Los Angeles and 126 seconds in San Jose, and fell by 131 seconds in Raleigh and 24 seconds in Detroit.
• In New York City, structural fires with the alarm source code CLASS-3 (read as alarm transmissions; the code is undefined in FDNY's dictionary) had a 90th percentile 98 seconds slower than phone calls in 2025.

Frequently asked questions

Related guides, services and research

• Fire alarm inspection requirements
• Fire sprinkler inspection requirements
• Office building fire inspection checklist
• Volunteer Fire Departments by State
• Fire Extinguisher Recalls and Smoke Alarm Recalls
• Fire Protection Acquisitions in the U.S., 2020 to 2026
• Hotel Fire Safety Compliance

Methodology

Selection: on 2026-10-27 we searched the Socrata Discovery API, ArcGIS Hub and city CKAN portals for U.S. city datasets with incident-level fire timestamps (a start stamp and a unit on-scene stamp). Eight qualified with 2025 records: New York City (FDNY Fire Incident Dispatch Data), San Francisco (Fire Department and EMS Dispatched Calls for Service), Los Angeles (LAFD Response Metrics Raw Data), San Jose (Fire Incidents), Cincinnati (Fire Incidents CAD), Detroit (Fire Incidents), Louisville Metro (Fire Incident) and Raleigh (Fire Incidents). Seattle publishes no arrival stamp; Baton Rouge publishes dispatch but not arrival; Austin publishes the metric only aggregated by response area and is used as a reference.

Inclusion: calendar 2025 (Louisville's file ends 2025-12-29). Fire incidents only where coded: FDNY Structural Fires group; San Francisco call type Structure Fire / Smoke in Building; NFIRS 111 Building fire in Detroit, Louisville and Raleigh; San Jose final category Structure Fire; Cincinnati structure fire dispatch codes (protocol 69, D and E levels). Los Angeles cannot be filtered to fires, so all Non-EMS emergency incidents are used. One record per incident: the earliest on-scene stamp among units (Louisville: engines, trucks, quints, rescues and chief cars, ignoring support apparatus; Detroit: exposure 0).

Exclusions: incidents with no usable start or on-scene stamp, zero or negative intervals, and intervals over 60 minutes (table t8). New York City uses FDNY's own validity flag. Los Angeles times are published as time of day without a date; intervals were computed modulo 24 hours. San Jose stamps are to the minute and its yearly files use different day-month order; each row was parsed with one consistent format.

Derived metrics: median and 90th percentile use linear interpolation between ranked values. Thresholds: 8 minutes from the Austin Fire Department's published goal (first unit within 8 minutes of phone pickup, 90% of the time); 240 seconds of travel for 90% of incidents from NFPA 1710 as summarized by the City of Santa Monica; 6 minutes as an unsourced sensitivity line only. Change compares the same method in 2019 (Louisville 2021). All calculations are in scripts/analyze.py and scripts/extras.py and reproduce every CSV.

Conflicts and checks: where a department publishes an average for the same interval, we compared it with our recomputed mean (table t7). New York City matched FDNY's fiscal 2025 average of 4:35 exactly; Los Angeles was within one second of LAFD's published Non-EMS operational average. We did not adjust any city's data.

Limitations: cities start the clock at different points, code incidents differently and record on-scene arrival by crew button press, so the ranking is indicative. Raleigh and Louisville have about 200 to 400 building fires a year, so their percentiles move easily. NFPA standards are cited by number and published objective only; adoption and enforcement are local, and nothing here is legal or operational advice.

Last updated 2026-10-27. Keyword check: neither "fire department response time" nor "average fire department response time" appears in the site's Ahrefs exports (alarm, door, inspection, suppression and damper seeds), so the plainest searcher phrasing was used.

This report presents published data with its limitations. It is not legal, engineering, financial or investment advice. Requirements depend on the code edition your state and local fire marshal have adopted; confirm with your authority having jurisdiction.

Sources, ranked by quality

Tier 1: primary and official sources (18)

• [1] Fire Department of the City of New York (FDNY) via NYC Open Data. Fire Incident Dispatch Data (dataset 8m42-w767). Incidents 2005-01-01 to 2026-09-30; 2019 and 2025 used. Accessed 2026-10-27.
• [2] FDNY via NYC Open Data. Fire Incident Dispatch Data: field descriptions (data dictionary attachment). Current dictionary. Accessed 2026-10-27.
• [3] NYC Mayor's Office of Operations. Mayor's Management Report, Fiscal 2025: Fire Department chapter. Fiscal years 2021 to 2025 (July to June). Accessed 2026-10-27.
• [4] San Francisco Department of Emergency Management and Fire Department via DataSF. Fire Department and Emergency Medical Services Dispatched Calls for Service (dataset nuek-vuh3). Calls received 2019 and 2025; refreshed daily. Accessed 2026-10-27.
• [5] Los Angeles Fire Department via Los Angeles Open Data. LAFD Response Metrics: Raw Data (dataset n44u-wxe4). Incident numbers for 2019 and 2025 (last update 2026-02-10). Accessed 2026-10-27.
• [6] Los Angeles Fire Department via Los Angeles Open Data. Citywide Response Metrics: Updated (dataset kcsj-s69p). Monthly averages, January 2019 to December 2025 used. Accessed 2026-10-27.
• [7] City of San Jose Fire Department via San Jose Open Data. San Jose Fire Incidents (yearly CSV files 2019 and 2025). Calendar years 2019 and 2025. Accessed 2026-10-27.
• [8] City of Cincinnati Fire Department via Cincinnati Open Data. Cincinnati Fire Incidents (CAD) including EMS: ALS/BLS (dataset vnsz-a3wp). Incidents created 2019 and 2025; refreshed daily. Accessed 2026-10-27.
• [9] City of Cincinnati via Cincinnati Open Data. Cincinnati Fire Incidents (CAD) Data Dictionary (PDF attachment). Current dictionary. Accessed 2026-10-27.
• [10] City of Cincinnati via Cincinnati Open Data. CFD Data Visualization Exclusions (PDF attachment). Current list. Accessed 2026-10-27.
• [11] City of Detroit Fire Department via Detroit Open Data Portal. Fire Incidents (ArcGIS feature layer). Incidents called 2016 to 2026; 2019 and 2025 used. Accessed 2026-10-27.
• [12] Louisville Metro Government via Louisville Open Data. Louisville Metro KY: Fire Incident (ArcGIS feature layer). Incidents 2021-01-01 to 2025-12-29; 2021 and 2025 used. Accessed 2026-10-27.
• [13] City of Raleigh Fire Department via Raleigh Open Data. Fire Incidents (ArcGIS feature layer). Incidents dispatched 2019 and 2025. Accessed 2026-10-27.
• [14] Austin Fire Department via Austin Open Data. AFD Standard of Coverage 2022-2024 (dataset 68my-ghq8). Calendar years 2022 to 2024. Accessed 2026-10-27.
• [15] City of Santa Monica. T12 Emergency Resp Times Fire (dataset pbfu-vhjg): description of the NFPA 1710 travel-time objective. Published 2015; describes the standard. Accessed 2026-10-27.
• [16] National Fire Protection Association (NFPA). NFPA 1710 standard development page. Last stand-alone edition 2020. Accessed 2026-10-27.
• [17] National Fire Protection Association (NFPA). NFPA 1750 standard development page. Current edition 2026. Accessed 2026-10-27.
• [18] FDNY via NYC Open Data. FDNY Monthly Response Times (Historical) (dataset j34j-vqvt). Designated historical; no longer updated. Accessed 2026-10-27.