← All articles

When Seattle Calls the Fire Department: A Year of 911 Data, Hour by Hour

Published October 13, 2026 · 9 minute read

Seattle's fire department runs on a daily rhythm you can see from the data alone. In the twelve months from October 2025 through September 2026, Seattle Fire logged 112,001 incidents, or about 307 a day. They did not arrive evenly. The quietest hour of the day, 4 a.m., produced 2,384 incidents over the whole year. The busiest, the 1 p.m. hour, produced 6,137. That is about 2.6 times as many, and the difference is mostly explained by one fact: the fire department is mostly a medical service, and medical emergencies happen when people are awake and moving around.

We pulled every incident in the city's real-time Seattle Fire 911 dataset for that period and counted them by hour, weekday, month and call type. Below is what the clock looks like, why some call types keep their own schedules, and what changed over the last four years.

The daily curve: a 4 a.m. trough and an afternoon plateau

Volume falls through the night, bottoms out in the 4 a.m. hour and starts climbing around 6 a.m. By 9 a.m. it is more than double the overnight low. From 11 a.m. to 5 p.m. it holds on a broad plateau, between about 5,880 and 6,140 incidents per hour-slot over the year, with the single highest hour at 1 p.m. After 6 p.m. the decline is gradual; even the 11 p.m. hour (4,036 incidents) is busier than any hour from midnight to 8 a.m.

Put another way: the seven hours from 11 a.m. through 5 p.m. make up 29 percent of the day but carried about 37.5 percent of all incidents. The six hours from midnight to 6 a.m. carried about 15.6 percent. On an average day that works out to roughly 16 or 17 incidents in the 1 p.m. hour across the whole city, and about six or seven in the 4 a.m. hour.

The reason is the call mix. "Aid Response," the department's basic medical call, accounted for 55,737 incidents, almost exactly half of the total. Add "Medic Response" (11,781, the paramedic-level call) and other medical types and the curve is dominated by medical emergencies. Falls, sudden illness and injuries track daytime activity: people commuting, working, shopping and visiting clinics and care facilities. Fires themselves are a small slice of the workload, a point we covered in our look at the most common 911 call types.

For residents, the practical takeaway is modest but real. The hours when the city is busiest are also the hours when more units are already committed to other calls. This dataset does not include response times, so it cannot tell you whether help arrives more slowly at 1 p.m. than at 4 a.m. What it does show is when the system is under the most simultaneous demand.

Weekdays are nearly flat; Sunday is the quiet one

Day of the week matters much less than hour of the day. Because the twelve-month window contains 53 Wednesdays and 52 of every other weekday, the fair comparison is the average per day rather than the raw total.

Thursday averaged about 316 incidents per day and Sunday about 292, a gap of roughly 8 percent. Monday through Friday all sit within a few percent of one another. Saturday is a little below average and Sunday about 5 percent below. That fits the same explanation as the daily curve: fewer people commuting and working, and fewer people in offices, schools and commercial buildings, means fewer medical calls and fewer building alarms. Whatever extra late-night calls weekend nightlife generates, they do not lift Saturday above an ordinary weekday.

Different call types keep different clocks

The overall curve hides call types with their own schedules. The dataset includes hour-by-hour counts for three that are worth comparing: automatic fire alarms, elevator rescues, and overdose responses. To compare shapes rather than sizes, the chart below shows each type's hourly share of its own yearly total, with the all-incident curve as a dashed reference line. The red dot marks each type's peak hour.

Automatic fire alarms follow business hours

Alarm calls (10,606 in the hourly series) broadly follow the business day, but they ramp up earlier and harder than calls overall. They rose from 226 in the 4 a.m. hour to 600 in the 9 a.m. hour, an increase of about 2.7 times, compared with about 2.2 times for all incidents over the same hours. Alarms peak in the 11 a.m. hour (634), two hours ahead of the overall peak, and about 49 percent fell between 9 a.m. and 6 p.m. That shape is consistent with alarms being triggered by activity inside occupied buildings: cooking, construction and maintenance work, testing, and people moving through commercial space during the workday. The data does not say which alarms were false, but the timing points to occupancy rather than to fires.

Elevator rescues peak at 5 p.m.

"Rescue Elevator" calls, people stuck in elevators, numbered 2,097, more than structure-fire dispatches under the "Fire in Building" type. They have the most extreme daily swing of any type in the data: 27 in the 3 a.m. hour against 152 in the 5 p.m. hour, a ratio of about 5.6 to 1. Volume nearly doubles between the 6 a.m. and 7 a.m. hours, stays elevated through the day and peaks as offices empty out: 152 in the 5 p.m. hour and 141 in the 6 p.m. hour. Elevator entrapments require someone to be in the elevator, so the curve is essentially a record of when Seattle's mid- and high-rise buildings are busiest. The peak lines up with the end of the workday, when the most people are riding down at once.

Overdose responses run late

"Medic Response- Overdose" is the one type in this set that does not follow the daytime pattern. Its busiest hour is 9 p.m. (75 incidents), followed by 11 p.m. (72) and 10 p.m. (68). About 17.8 percent of overdose responses fell between 9 p.m. and midnight, compared with about 12.1 percent of all incidents. It is also comparatively busy after midnight: about 20 percent of overdose responses fell between midnight and 6 a.m., against 15.6 percent overall. Its quietest hours are 5 and 6 a.m.

Two cautions apply. First, the numbers are small. With 1,211 incidents spread over 24 hours, a single hour holds only 25 to 75 calls, so hour-to-hour wiggles are noise; the evening lean is the signal. Second, this counts only incidents dispatched under the overdose call type. Some overdoses are dispatched as general aid or medic responses when the caller does not describe a drug emergency, so this is not a count of all overdoses in the city.

The four-year trend: volume fell after 2024

The dataset reaches back years, which lets us place the latest twelve months in context. Monthly totals from January 2022 through September 2026 are below; the most recent twelve months are in amber.

Annual volume rose from 119,931 in 2022 to 125,678 in 2023 and 129,629 in 2024. It then fell to 119,299 in 2025. The most recent twelve months (October 2025 through September 2026) totaled 112,001, which is 9.8 percent fewer than the 124,142 incidents in the twelve months before (October 2024 through September 2025) and 13.6 percent below calendar 2024.

The timing of the drop is specific. Month-over-same-month-last-year comparisons turned clearly negative around June 2025, when the count was about 12 percent below June 2024. The gap was widest from November 2025 through April 2026, with each month running 14 to 19 percent below the year before. It narrowed in May 2026 (about 9 percent lower), and from June through September 2026 the counts were roughly level with 2025: June up about 1 percent, July down about 4 percent, August up about 3 percent and September up about 4 percent. So the decline looks like a step down that took place over roughly a year, not a steady slide that is still going on.

What the data can and cannot tell us about why

A count of dispatched incidents can change because emergencies changed, because more people called or fewer did, or because the way calls are handled and categorized changed. This dataset alone cannot separate those. A few possibilities are worth naming, without claiming any of them is the answer:

• Triage and categorization. The past year's data includes call types such as "Low Acuity Response" (1,682 incidents) and "Low Acuity Referral" (79). If more lower-urgency calls are handled through a different pathway, or referred elsewhere without a fire unit being dispatched, the published incident count could fall without any change in how often people need help. We have not confirmed when these types were introduced or how they are used.
• Real changes in demand. Medical calls respond to things like weather, public health conditions and how many people are downtown. A sustained change in any of these could move the total.
• Data completeness. The real-time feed is published for public use, and records can be added or reclassified after the fact. A count pulled later could differ slightly.

Separating these would take information from the department itself, such as unit-level workload or call-handling policy changes. We would rather flag the question than guess at the answer.

Reading the live feed with the clock in mind

If you follow the live feed or our daily recap, these patterns help make sense of what you see. A busy early afternoon is normal. A cluster of alarm activations mid-morning is normal. A run of elevator rescues downtown around 5 p.m. is normal. Activity that stands out from the curve, such as a surge of fire calls at 3 a.m., is more likely to be worth a closer look. Our guides to Seattle Fire response codes and how Seattle's 911 dispatch works explain what each incident type means and how calls get routed.

← All articles