The Crash Indicator Treats a Fender-Bender and a Fatality the Same. Your Vetting Shouldn't.

About two years ago I had a situation I still think about when I explain this.

Two carriers in my approved pool. Both sat in the low 70s on the Crash Indicator BASIC — not ideal, but below the 80% intervention threshold. Both had two crashes in the prior 24 months. On paper: identical risk.

Carrier A had backed into a guardrail at a truck stop. Tow-away, no injuries, driver cited for improper backing. Minor property damage. Second crash: clipped a barrier on a construction zone entrance. Nobody hurt.

Carrier B had a crash on I-65 in Kentucky. Loaded tanker, southbound, rear-ended a stopped passenger vehicle. Driver and the front-seat passenger were hospitalized. Second crash: wet highway, tractor-trailer crossed the centerline, sideswipe with a box truck. Both drivers treated at the scene.

Same score. Different universe. I tendered loads to Carrier B twice before I actually looked at what was in the underlying crash report. I got lucky — both loads were clean. But if Carrier B had had a bad event on my freight with that history behind them, I'd have been in a very difficult position trying to explain in discovery why I never looked past the percentile.

The Crash Indicator BASIC percentile is a useful starting signal. It is not the whole picture. Here's how to read what's behind it.

What the Score Actually Counts

The Crash Indicator is built from "reportable accident" data that motor carriers must report and that state police forward to FMCSA's MCMIS system. Under 49 CFR § 390.5, a "reportable accident" means a fatality, an injury requiring immediate medical treatment away from the scene, or property damage requiring a vehicle to be towed. That's the threshold.

A parking lot fender-bender that sends a truck to a body shop: reportable. A highway rollover with three fatalities: also reportable. Both count as 1.

The BASIC does apply time-weighting — more recent crashes count more than older ones — and it normalizes by the carrier's inspection count, which introduces its own distortions. But within the pool of crashes that qualify, severity doesn't factor into the score. A tow-away is a tow-away.

That's by design. FMCSA built the BASICs to predict crash frequency for intervention purposes. Not to calculate a broker's civil liability exposure. Those are different goals. For a freight broker trying to assess litigation risk after Montgomery v. Caribe Transport II, LLC, crash severity matters a great deal.

Where the Actual Crash Data Lives

Your starting point is the FMCSA Company Snapshot on SAFER. Pull the carrier, scroll to "Crashes." SAFER shows the 24-month total broken down by fatal, injury, and tow-away.

That breakdown is the first useful cut. Fatal crashes are obvious red flags regardless of anything else. Injury crashes — even one with documented spinal or traumatic brain injury — change the calculus entirely. Tow-aways with no injuries are a different animal.

For more detail, pull the Crash Detail through the Safety Measurement System on the FMCSA portal. Crash reports vary in how much narrative they carry, but you can usually get crash date, location, highway type, and any violation cited to the driver. Many reports include KABCO severity codes: K (killed), A (incapacitating injury), B (non-incapacitating injury), C (possible injury), O (no apparent injury).

KABCO A crashes should get your full attention above everything else. A crash where someone walked away incapacitated — spinal fracture, traumatic brain injury, loss of limb — is the kind of event that generates eight-figure verdicts. If a carrier has a KABCO A in their 24-month window, you need to know it before you tender.

What to Look For Beyond the Severity Code

The KABCO code tells you how bad the crash was for the people in it. These other details tell you what kind of operation you're actually dealing with.

Highway type. A parking-lot fender-bender and a highway crash at 65 mph carry different risk profiles even if both are injury-coded. High-speed, high-consequence environments tell you about the carrier's OTR operation. Local delivery crashes tell you about urban maneuvering. If you're putting this carrier on an interstate lane, the interstate crashes in their history are what's relevant.

Load type at time of crash. If a carrier you're considering for a high-value or heavy-haul load had their crashes while running similar freight, that's directly applicable. If their history is entirely local light-freight runs, you're vetting a different operation than what you're asking them to do.

Whether the driver got cited. A crash where the driver received a violation is documented evidence of fault at the scene. Two crashes with driver citations — especially under Part 392 (driving rules) or Part 395 (hours of service) — tells you something about driver qualification and carrier supervision. 49 CFR § 391.11 requires carriers to verify that drivers are qualified. When a driver gets cited, stays behind the wheel, and gets cited again, that progression matters. Plaintiff's attorneys build negligent-entrustment arguments on exactly that pattern.

Time of day. FMCSA crash reports don't always make this easy to extract, but a crash at 2:30 AM on a long-haul run is a fatigue signal. Cross-reference with the Hours of Service BASIC (Part 395). A carrier with both HOS violations and a late-night crash in their history is a different risk than a carrier with a daytime parking lot incident.

Preventability Is a Separate Question

I see brokers conflate these constantly. Preventability — coded as P, NP, or U (preventable, non-preventable, unknown) in SMS — asks whether the driver could have avoided the crash with reasonable care. Severity asks how bad it was.

A non-preventable crash can still kill someone. A carrier can have two NP crashes — a deer strike and a driver who blew a red light into them — and a perfect operational record. High severity, not the carrier's fault.

The problem is that preventability coding in SMS is messy in practice. A significant portion of crashes show as U (unknown), meaning FMCSA hasn't made a determination. The Challenge process, which lets carriers contest preventability, is backlogged. Don't treat P/NP codes as your primary severity signal. Go to the underlying facts.

After Montgomery, This Analysis Isn't Optional

Before May 14, you could credibly argue that FAAAA preemption shielded brokers from state negligent-selection claims. The Supreme Court ended that argument unanimously in Montgomery v. Caribe Transport II, LLC. The decision is in. Brokers can be sued in state court for the carrier choices they make.

What that means in practice: when you tender a load to a carrier that later crashes and injures someone, the plaintiff's attorney will subpoena your vetting file and then work backward through the carrier's history. If that carrier had a KABCO A crash 14 months ago and you never looked past the percentile, that's going to come out in discovery. The question — "did you know, or should you have known, that this carrier had previously put someone in the hospital?" — is now a state-law negligence question for a jury to answer.

I got lucky with Carrier B. In the post-Montgomery environment, luck is not a risk management strategy.

Walk Through a Concrete Example

Consider MC-1629847, DOT-4388201 — a carrier I'm constructing to make this specific. Active authority for three years. Crash Indicator at 68th percentile. Two crashes in the 24-month window.

Crash one: 18 months ago. Distribution center parking facility. Driver backed into a dock door, trailer sustained $8,000 in damage, no injuries, driver cited for failure to use a spotter. Tow-away. Coded preventable.

Crash two: 7 months ago. US-30 in Indiana. Loaded reefer, 44,000 lbs. Rear-end collision with a passenger car that had slowed for a work zone. Front passenger in the car sustained a cervical spinal fracture — KABCO A, incapacitating injury. Driver had 9 hours and 42 minutes on duty prior to the event. No hours-of-service violation cited, but that number is in the report. Coded unknown preventability.

68th percentile. The same score you'd see on a dozen carriers in your approved list this week. Completely different litigation profile.

If that carrier rears into someone on your freight and the plaintiff's attorney finds this prior KABCO A crash — seven months ago, same load type, same kind of highway, similar duty hours — you are going to spend a very uncomfortable day in deposition explaining why you only pulled the BASIC score.

The annual cargo value on a decent reefer lane runs $300,000 to $500,000 per load. The BIPD minimum under 49 CFR § 387.9 is $750,000. A serious spinal injury verdict in Indiana can exceed both by multiples. The cost of pulling the underlying crash report is thirty seconds.

How I Document This

Every carrier load tender in my workflow now includes a crash history note. Not just the BASIC percentile — the underlying breakdown and any notable crashes.

Standard entry: "Crash Indicator: [X]th percentile. [N] crashes per SAFER in the last 24 months: [X] fatal, [X] injury, [X] tow-away. Crash detail reviewed [date]. [Brief description of any notable crash: severity code, road type, load type, driver violation if cited.] Screenshot attached."

If there are zero crashes, that goes in the note too: "Zero reportable crashes in 24-month window per SAFER as of [date]."

If I see a KABCO A or K crash in the history and I tender anyway, the note explains my reasoning. Maybe the crash was clearly non-preventable by the evidence in the report. Maybe the carrier has had a major turnover in their driver pool since then. Whatever the reason, it's in the record. That reasoning — documented before the load moves, not reconstructed afterward — is what matters if the carrier later has an event on my freight.

Most brokers don't do this. They pull the score, note that it's under 80, and move on. That's not vetting crash history. That's confirming a number. After Montgomery, there's no legitimate argument for not spending two extra minutes to know what's actually behind it.

— Mason Lavallet

Founder, DOTScreener.com