Before the Call — I-35W Bridge Collapse
At 6:05 in the evening on August 1, 2007, the central span of the I-35W bridge — a major interstate crossing over the Mississippi River in the heart of Minneapolis — failed and fell. It was rush hour. The bridge carried 141,000 vehicles a day. In the seconds before collapse, roughly 111 vehicles were on the deck, along with a school bus from a North Minneapolis recreation program carrying 52 children and a crew of highway maintenance workers resurfacing the roadway.
The bridge fell approximately 64 feet into and alongside the Mississippi. The central span hit the water. Other sections landed on the riverbanks and on a rail yard below. The collapse took about ten seconds. Thirteen people died. One hundred forty-five were injured. The school bus lodged on a collapsed section above the waterline — all 52 children survived, many of them evacuated by construction workers from a nearby site who arrived before any fire or EMS units.
The I-35W collapse shares a category with the 2024 Francis Scott Key Bridge collapse in Baltimore — a major bridge, a major urban waterway, vehicles in the water, multi-agency response. But the two incidents are meaningfully different in ways that matter for dispatch training. The Key Bridge fell at 1:28 in the morning with very limited traffic. The I-35W fell at 6:05 in the evening during rush hour with over a hundred vehicles. The Key Bridge collapse was preceded by a distress call from the container ship that gave police 90 seconds to stop traffic. The I-35W collapse had no warning at all — nothing preceded it, no structural alarm, no evacuation, no radio traffic. In one moment, the bridge was there. In the next, it was gone.
What the I-35W teaches dispatch is distinct from the Key Bridge: the problem of assembling a mass casualty incident from fragmented first calls, the civilian rescue window before units arrive, and the specific complexity of simultaneous water rescue and land-based MCI operations sharing a single incident command structure.
The I-35W collapse generated 911 calls within seconds — among the fastest call-to-dispatch timelines for a mass casualty event on record. The challenge isn't that the callers were unhelpful. It's that each of them saw a piece of the same catastrophic event from a different angle, and none of them had ready words for what they were describing. "The bridge fell" is simple. Getting there, in the first thirty seconds of the first call, from a screaming witness on a riverbank, is not.
- The first call is a signal, not a full picture. For an event this large, the first 911 call will almost never contain enough information to fully characterize the incident. What it will contain is enough to start moving resources. A caller saying "the bridge collapsed into the river, cars are in the water" is sufficient to dispatch a large initial assignment and begin escalation — you don't need to know how many cars or how far they fell before you move.
- Corroboration comes fast on a high-visibility incident. The I-35W bridge crossed the Mississippi in the middle of a major city at rush hour. Within a minute of the collapse, calls were arriving from multiple vantage points. Each call either confirms, expands, or contradicts the developing picture. A dispatcher tracking corroboration — "third caller now confirms vehicles in the water" — is building confidence in the dispatch decision faster than any single call can provide.
- You can dispatch large before you have details. The consequence of under-dispatching a bridge collapse is catastrophic. The consequence of over-dispatching one is a longer debrief. On any report of bridge failure with vehicles involved, the appropriate initial response is a mass casualty assignment: multiple engines, truck companies, water rescue, EMS, supervisors. The specifics get added as calls come in. The initial dispatch decision doesn't wait for them.
- The queue problem at scale. In the first minutes, the PSAP is receiving simultaneous calls describing the same event. Each call adds information — but each call also pulls a call-taker. The dispatch supervisor's job in that window is triaging inbound volume: which calls are adding new actionable information, which are duplicates, and whether additional staffing needs to come online immediately.
The civilian rescue window at I-35W is one of the most significant pre-arrival self-rescue events in American mass casualty history. The construction workers who evacuated the school bus and the bystanders who waded into the river weren't waiting for permission — they responded to what they saw. That's not unusual in the first minutes of a sudden catastrophic event. What matters for dispatch is understanding what it can do to support that window rather than simply waiting for units to arrive.
- Callers who are actively rescuing are a resource, not just a reporting source. A caller who says "I'm in the water, I've got someone, I need help" is giving you their location, their status, and what they need. They are an asset. Giving that caller information — "units are two minutes out, stay where you are, keep the victim's head above water" — keeps them effective longer and gives arriving units a known rescue location to target first.
- Communicating civilian rescue activity to incoming units before they arrive. Units arriving to a mass casualty scene in the first minutes are making rapid decisions about where to go and what to prioritize. Dispatch relaying "multiple civilians are in the water on the east bank assisting survivors" gives arriving units actionable information. It prevents duplication of effort where a unit drives past an active civilian rescue without knowing it's happening.
- The school bus is a specific accountability challenge. A school bus with 52 children represents a known occupancy at a specific location — all 52 names are on a roster. Dispatch coordinating with the school district or recreation program to get that roster to incident command is a support function that can happen from the console while ground operations run. Knowing who was on that bus, whether all 52 are accounted for, and where they're being transported is a separate accountability track from the general MCI operation.
- Civilian rescuers can become victims. Bystanders in the water, on unstable collapsed sections, or near submerged vehicles are themselves at risk. Dispatch passing this concern to arriving incident commanders — "multiple civilians in the water performing rescues" — allows IC to immediately assign personnel to evaluate civilian rescuer safety alongside victim rescue. The first responder's job includes not creating additional casualties.
The I-35W collapse is architecturally two incidents that share a location and a command structure. The Mississippi River water rescue operation — dive teams, swift water resources, Coast Guard, boat operations — runs on different protocols, uses different equipment, and talks on different channels than the land-based extrication and triage operation on the collapsed sections. Merging them into a single undifferentiated resource pool creates accountability gaps and delays the right resources reaching the right problems.
- Geographic sectors need to be established early and communicated to dispatch. At an incident this complex, IC will typically divide the scene into sectors — in this case, a water sector, a north bank sector, a south bank sector, and potentially a rail yard sector where debris landed. Dispatch tracking which resources are assigned to which sector, and routing incoming resources to the right IC branch, is a core function that breaks down if sectors aren't clearly defined and relayed back to the console.
- Water rescue resources have different request and staging protocols than land resources. Requesting a dive team is not the same as requesting an engine company. Coast Guard coordination runs through different channels. Boat resources may be staging at a marina rather than a street address. At I-35W, the Mississippi is federal water with Coast Guard jurisdiction, but city fire also has water rescue capability. Both were operating. Knowing who to call for what, before the incident, is the difference between a two-minute resource request and a fifteen-minute one.
- Simultaneous triage and water rescue create competing resource demands. EMS units triaging patients on the bridge spans cannot simultaneously be staging for water rescue retrieval. Dispatch managing the resource board needs to track which units are committed to triage and which are available for transport or water-side staging. As the first wave of survivors is triaged and transported, the resource picture shifts. Dispatch updating available resources in real time prevents IC from requesting units already committed elsewhere.
- Hospital notification needs to happen early and at volume. One hundred forty-five injured in a single incident is hospital-system-level stress. Dispatch or a dedicated medical coordinator needs to begin hospital notifications within the first minutes — not waiting for triage to finish before alerting trauma centers. Coordinating that wave of hospital alerts from dispatch, rather than each transporting EMS unit calling independently, creates a more coherent medical system picture.
The Key Bridge and the I-35W are the same incident type — major bridge, navigable water, catastrophic structural failure — but they are fundamentally different dispatch events. Understanding the difference matters because the no-warning scenario is the more common historical pattern, and the warning scenario is the rare exception.
- With warning: dispatch is a prevention operation. At Key Bridge, the Dali's distress call gave dispatch a 90-second window. Police stopped traffic. The response begins before the MCI. Dispatch's job is executing a rapid bridge closure: calling MTA Police, coordinating the traffic stop, escalating to incident command before the collapse. Most of the life-safety work happens in that window.
- Without warning: dispatch is an assembly operation. At I-35W, there was no warning. Dispatch receives the first calls seconds after the event, not before it. The job becomes assembling an accurate picture from simultaneous fragmented calls, making the initial dispatch decision on incomplete information, and escalating rapidly from "possible major incident" to "confirmed mass casualty event." The response begins after the MCI has already started.
- The accountability problem is worse without warning. At Key Bridge, first responders knew roughly who was on the bridge — a documented construction crew. At I-35W, 111 vehicles at rush hour means no immediate accountability baseline. Who was on the bridge is unknown until the response builds a picture from survivors, witnesses, vehicle registrations, and eventually recovered victims. That gap drives the recovery operation for thirteen days. Dispatch's role is to document everything — every caller who says "I saw a red sedan go off the south side" — because that information may matter days later.
- What I-35W teaches that Key Bridge doesn't: the unwarned collapse. Most infrastructure failures don't send a mayday first. The I-35W bridge fell because a steel gusset plate undersized since original construction finally gave way under construction load added the day of the collapse. There was no signal, no alarm, no pre-event radio traffic. Your first awareness of the event is the calls. Your job is to move from those calls to a coherent mass casualty dispatch as fast as the information allows.
✍️ Your Reflection
Complete this section and print your response — or save a PDF to share with your supervisor.