Most lights-and-siren runs buy seconds that never change the outcome.

What the evidence shows

The field has adopted the measure that matters

The 2022 Joint Statement signed by fourteen national EMS and public-safety organizations states the standard plainly: lights and siren "should only be used for situations where the time saved by L&S operations is anticipated to be clinically important to a patient's outcome."¹ That is a benefit test, set by the profession itself, and the U.S. Fire Administration has published aligned guidance.⁸ The honest question for any department is simply how much of its current siren use passes its own field's test.

For a small, identifiable set of calls the test is passed easily: cardiac arrest, severe trauma, and similar events where minutes genuinely move survival. Nobody disputes those runs. The operational problem has always been knowing, in advance, which call is which — and that is precisely what modern dispatch triage addresses. The systems profiled below hold lights-and-siren for the calls their 911 centers identify as time-critical, and run the rest quiet.

The time saved is small — and rarely the deciding factor

The federal EMS literature review found lights-and-siren shortens response intervals by 1.7 to 3.6 minutes and transport intervals by 0.7 to 3.8 minutes, across urban, suburban, and rural settings⁵; the Joint Statement's summary of more than a dozen studies puts the average saving at 42 seconds to 3.8 minutes.¹ For most calls that margin does not change the clinical picture: in the six-million-call national analysis, 6.9% of 911 EMS calls involved a potentially life-saving intervention, under a deliberately generous definition.³ The EMS systems now reforming their own protocols reach the same assessment — Wake County's EMS director puts the genuinely time-critical share of calls at about one in five.¹⁸

The crash risk is modest per run — and almost entirely gratuitous

In the national analysis of roughly 19 million 911 responses, crash rates were about 50% higher (adjusted) with lights-and-siren during the response phase — 5.4 versus 4.6 crashes per 100,000 responses — and nearly two and a half times higher during transport: 17.1 versus 7.0 per 100,000.⁴ Stated honestly, the per-run odds are small — which is exactly why the risk rarely registers in any single department's experience. Two things change the picture. The first is volume: across millions of responses a year, small per-run rates become a steady national toll. The second is what department statistics never capture — crashes among civilian drivers reacting to a siren that never touch the emergency vehicle and never enter its records. And because the seconds purchased rarely matter clinically, nearly all of this risk is taken in exchange for nothing.

What happens when departments stop taking it? In the national quality collaborative that ran from 2022 to 2024, the nine systems reporting aggregate data — 738,099 responses — cut lights-and-siren response use from 73% to 47% of calls.² Agencies that earlier reduced lights-and-siren response to 20–33% of calls did so, in the Joint Statement's words, "without any discernable harmful effect on patient outcome."¹

The noise itself is a health cost

Beyond crashes, the warning sound is a repeated exposure for the neighborhoods it passes through. European public-health agencies now attribute 66,000 premature deaths a year across Europe to chronic transportation-noise exposure, through stress-hormone and blood-pressure pathways.⁹¹⁰ Those figures measure chronic ambient noise, not sirens specifically — but sirens are an acute, repeated form of noise exposure operating through the same physiological pathways, concentrated on residents near busy response corridors and stations.

And the warning often arrives too late to work

A closed-road field demonstration by a New Jersey EMS cooperative found that an approaching siren first became audible to a driver in a closed, occupied car — windows up, radio on, driver on a phone — at about 280 feet, roughly seven seconds of warning at the ambulance's 25-mph approach speed.¹¹ The intuition that "drivers hear sirens just fine" does not survive contact with modern, well-insulated vehicles. Much of the risk a siren creates is not offset by the warning it provides.

The lights are not the controversy

It is worth separating the two warning devices bundled in "L&S." The noise burden, the sleep disruption, and the audibility failure all attach to the siren; emergency lights carry none of that cost, and lights-only response modes exist in department policies today. The crash research studies lights and siren as a package, so reduced crash figures cannot be attributed to the siren alone — but every other cost on this page can be. A city that never touched its lights policy and only tightened when sirens sound would capture the entire noise-side benefit.

The reform runs through existing law — without putting the choice on the driver

In most US jurisdictions the traffic code permits emergency vehicles to use lights and siren; it does not require it — and the permission is typically framed as use "when reasonably necessary," language that speaks to moments (an intersection, a blocked lane), not to entire trips run loud from door to door. The decision is discretionary, which is what makes reform tractable: tightening when the devices are used needs no new statute, no labor renegotiation, and in many cities no council action. But discretion exercised one call at a time, by an operator already driving under pressure, is the wrong place to locate it. It is a judgment most responders neither want nor should have to make in the moment. The durable reforms move the decision off the individual and into written policy and dispatch.

The loud default is itself a policy artifact, not a law of nature. In Seattle — offered here as a documented example — the fire department's most recent publicly posted operating guidelines state that all responses "should be code red," meaning lights and sirens, "unless otherwise designated" by the dispatch center.²⁴ A default written by administrative document can be rewritten by the same instrument. That is the entire mechanics of this reform.

The most effective programs make the quiet response the default and the lights-and-siren response a defined exception, reserved by protocol for a narrow set of genuinely time-critical calls. Wake County is implementing exactly this: its 911 center sorts calls by reported symptom and acuity, and lights-and-siren is held for the roughly one in five calls that are time-critical — cardiac arrest, severe collisions, shootings.¹⁷¹⁸ The crew's job does not change. The protocol, not the driver, decides how they roll.

Three levers do the work, and they reinforce each other:

• A default-quiet policy. Make non-lights-and-siren response the standard and reserve Code 3 for a defined list of critical situations — the St. Louis model. The decision lives in the policy, not in a split-second judgment call.
• Dispatch-side screening. Set the response mode at the 911 center, before a unit rolls, from the caller's reported symptoms — so the call type, not reflex, governs the response.
• Substitution for lower-acuity calls. Route the large share of non-emergency 911 volume — much of it medical and behavioral — to mobile-integrated-health or alternative-response teams that never needed a siren in the first place.

Framed this way, the reform reduces what it asks of front-line responders rather than adding to it. It also narrows the liability that lights-and-siren crashes create for agencies and local governments — a point EMS-industry counsel makes explicitly.²¹ It does not ask a department to do less for the public — it reserves the warning devices for the calls where seconds genuinely change outcomes, which is the standard the field has already set for itself.¹