Two Crashes, Two Totally Different Carriers: What the FMCSA Crash Counter Misses

Two carriers. Both show two crashes in the last 24 months on SAFER. One of them I'd use tomorrow. The other I'd never tender a load to. The number didn't tell me which was which — the context did.

That context is the part most brokers skip. They see "2 crashes," run a mental checklist — BASIC percentile, authority age, OOS rate — and move on. That's almost always a mistake.

What FMCSA Actually Counts

Let's start with the definition, because it's narrower than most people think. Under 49 CFR § 390.5, a "recordable accident" for federal purposes is one that involves a fatality, a bodily injury requiring immediate medical treatment away from the scene, or disabling damage that requires a vehicle to be towed. That's it. A driver who rear-ends a parked car at 5 mph in a shipping yard, walks away fine, the truck drives off — that doesn't appear on SAFER. A driver who sideswiped a guard rail on the interstate, nobody hurt, truck limped off under its own power — also not there.

What that means is that SAFER's crash section is already filtered. What's in the count tends to be serious events. But the word "tends" is doing a lot of work, and that's where you have to read carefully.

The database tracks crashes from state-reported data — state police crash reports submitted to FMCSA. A carrier with five trucks running one region might have state-level reporting gaps. A large carrier running 48 states has near-complete data. Fleet size matters when you're interpreting the raw count. Two crashes on a fleet of four trucks operating locally is a meaningfully different signal than two crashes on a 200-truck national operation.

The Two Carriers I Mentioned

The first one — I'll call it Summit Ridge Freight, MC-1247893, DOT-3891047, out of Chattanooga, authority age about four years. Two crashes in 24 months, Crash Indicator percentile in the upper 40s. When I pulled the full SAFER snapshot, the inspections section showed 91 inspections over that same window. No OOS orders. The crash dates were 18 months apart. No compliance review had been triggered. Their MCS-150 was current. Nothing in the regulatory action section.

Ninety-one inspections and two reportable crashes over two years on a 12-truck operation. That's not a clean safety culture, but it's not a carrier with a systemic problem either. The crashes may well have been unavoidable events — a rear-end in heavy traffic, an intersection incident the driver didn't cause. FMCSA doesn't determine fault. The crash goes on the record regardless.

The second carrier — Keystone Prime Logistics, MC-1388204, DOT-4021765, authority age 14 months. Two crashes in the last 18 months. Crash Indicator percentile in the upper 70s. When I dug into the SAFER snapshot: 11 total inspections, 3 of them generating OOS violations. No compliance review, but the short authority age and low inspection count meant the percentile calculation was working off a thin denominator. And here's the part that actually got my attention — both crashes occurred within 60 days of each other.

Two crashes in 60 days on a carrier with 14 months of authority and a thin inspection record. That's a completely different profile than two crashes spread across 24 months on an established carrier with 91 clean inspections. But if all you looked at was the crash count, you'd see "2" on both carriers and apply the same weight.

Why the Percentile Can Mislead You

The Crash Indicator BASIC percentile is supposed to normalize for exposure — specifically, miles traveled or number of power units. FMCSA uses estimated miles from the MCS-150 for that calculation. The problem is that MCS-150 data is self-reported and often inaccurate. Carriers file it every two years, and some operators let it drift.

I've seen carriers with wildly understated mile estimates that show inflated Crash Indicator percentiles simply because their denominator is wrong. And I've seen the reverse — a carrier who updated their MCS-150 to reflect a recent expansion showing a lower percentile than their underlying record warrants.

None of this means the BASIC is useless. A percentile in the 80s is a real signal. But a percentile in the 50s or 60s doesn't let you off the hook for reading the underlying data. The raw numbers — crash count, inspection count, time between crashes, whether OOS orders followed — tell you more than the normalized score when you're looking at a smaller carrier.

What Follows a Crash Matters More Than the Crash Itself

The single most useful signal in the crash section isn't the count. It's whether regulatory action followed.

49 CFR Part 385 gives FMCSA the authority to initiate a compliance review (CR) and issue a safety fitness determination after an accident. When a serious crash — fatality, multiple injuries — occurs, state police typically flag it to the FMCSA regional office. A compliance review can follow. That review, and any resulting safety rating change (Conditional or Unsatisfactory), shows up in SAFER's ratings section.

If a carrier had a fatal crash and their rating didn't change — check the date. If the crash happened 18 months ago and they're still carrying "Not Rated" status, that might mean the compliance review is still pending, or it might mean the regional office hasn't initiated one yet. Either way, the absence of a rating change after a fatal crash is not confirmation that everything is fine. It's confirmation that you need to ask more questions.

If the crash happened and the carrier took an emergency OOS order under Part 385.13, that's in SAFER. That is a hard stop. A carrier who got put out of service for an imminent hazard and is now trying to get back into your routing guide deserves a long conversation, documented, before you tender anything.

The Broker's Exposure Post-Montgomery

After the Supreme Court's decision in Montgomery v. Caribe Transport II this past May, state-law negligent-selection claims against brokers are no longer preempted by the FAAAA. That ruling changed the discovery math.

Before Montgomery, the preemption argument gave brokers a reasonable legal shield. That shield is gone in state court. What's left is your file.

Here's how plaintiff's counsel works a crash case against a broker. They subpoena your carrier selection records. They look at what you knew and when you knew it. If the carrier you used had a visible Crash Indicator percentile in the 70s, a recent compliance review, and an OOS order on their record — and your screening notes say "checked SAFER, good to go" — that's a problem. You checked the box. You didn't read the record.

If the carrier had two crashes in the last six months on a 14-month-old authority and you tendered a $180,000 high-value load anyway with no additional documentation of why you made that call, that's a harder conversation than if you had a written justification in the file: why you assessed the risk, what you looked at beyond the surface numbers, what additional verification you did.

The number on the screen isn't your defense. Your documented reasoning is your defense.

When a New Crash Shows Up Between Screens

This is a scenario I didn't think about carefully enough early in my career. You screen a carrier in February. They pass. You add them to your routing guide. In April, a new crash gets added to their SAFER record — reportable, Crash Indicator percentile now over 75. You don't know. You tender a load in May. That load ends badly.

If you re-screened in May, you'd have seen it. You didn't, because once a carrier made the approved list, nobody was going back to check.

That's exactly what Continuous Monitoring in DOTScreener is built to catch. When a new crash record appears on a carrier you're monitoring, the system flags it and prompts a re-run of the carrier screen. The updated Carrier Selection Record gets filed with the timestamp, documenting that you were watching. If there's an adverse event and a plaintiff's lawyer wants to know why you kept using this carrier after a second crash appeared — your file shows you ran the re-screen and made a documented call.

It doesn't make the decision for you. It makes sure you're not the broker who didn't notice.

How I Document This

When I'm working a carrier with any crash history, my file for that carrier includes:

• The date I pulled the SAFER snapshot and what the crash section showed (count, dates, percentile)
• The inspection count and OOS rate from the same pull — I want the ratio visible in the file, not just the crash count
• Whether any OOS orders or compliance reviews followed the crashes, and what came out of them
• If crash dates are clustered within a short window, a note on why I still approved (or didn't) — the reasoning has to be in writing
• For newer authority carriers with any crash history: a note on fleet size and whether the percentile denominator is reliable

If the answer is "I looked at the crash section and the carrier still passes my screen," I want that written out with the basis for the conclusion — not just a checkbox that says "screened."

Crashes happen. Carriers with crashes aren't automatically carriers you can't use. But the brokers who get into trouble aren't the ones who used a carrier with a crash record — they're the ones who didn't read it, didn't document it, and couldn't explain their decision afterward.

That distinction is what survives discovery.

— Mason Lavallet

Founder, DOTScreener.com