Reading the Crash File: Why 'Non-Preventable' Is the Most Misunderstood Field on a Carrier's Record

Two carriers walk into a load board. Both have four DOT-reportable crashes in the last 24 months. Both have insurance, both have active authority, both have BASIC percentiles in roughly the same neighborhood.

Carrier A: of those four crashes, three are classified "Non-Preventable" by FMCSA's Crash Preventability Determination Program. The carrier's driver was rear-ended at a red light, hit by a wrong-way driver on an entrance ramp, and struck while legally parked. The fourth crash is "Undecided" pending review.

Carrier B: of those four crashes, three are classified "Preventable." Following too close, failure to maintain lane, and one rollover at a known-grade highway off-ramp where speed was a factor. The fourth crash is "Preventable" with a fatal outcome.

These two carriers have the same crash COUNT on their public file. They are not even close to the same risk profile. The number "4" alone tells you nothing.

Brokers who look at "Crashes (24mo): 4" and call it a day are missing the part of the record that actually matters — and post-Montgomery, that's the part the plaintiff's lawyer is going to point at.

What the Crash Preventability Determination Program is

The CPDP started in 2017 as a FMCSA pilot and became permanent in 2020. The idea: many crashes carriers were getting "credit" for in their CSA scores were crashes the carrier had no realistic ability to prevent — getting rear-ended by an inattentive four-wheeler, getting hit by a drunk driver crossing the median, getting struck while parked. The CSA system was treating these the same as a fatigued-driver-failed-to-yield crash. That's nonsense as a safety signal.

Under CPDP, carriers can submit a Request for Data Review through the FMCSA DataQs system asking the agency to designate a crash as Non-Preventable. The crash stays on the carrier's record — it's not deleted — but it's flagged "Non-Preventable" and is **excluded from the Crash Indicator BASIC calculation**. The crash itself is also visible to anyone (including brokers) doing diligence on the carrier.

There are now sixteen eligible crash types under CPDP — rear-end victim, struck by wrong-way driver, struck while legally parked, struck by suicidal individual, struck during commission of a crime, animal strike, infrastructure failure, several others. The full list is published by FMCSA.

Why this matters for the broker's diligence file

Two reasons.

One: the number "4 crashes" is a worse signal than "3 preventable, 1 non-preventable." A carrier whose crash file is mostly "preventable" is a carrier whose drivers are causing crashes. A carrier whose file is mostly "non-preventable" is a carrier that's been unlucky — they keep getting hit. Those are not the same level of operational risk, and they're not the same level of broker-side exposure.

Two: the Crash Indicator BASIC percentile is calculated AFTER non-preventables are stripped out. Which means a carrier with a 78% Crash Indicator percentile (alert) is at 78% AFTER the agency has already given them credit for the non-preventable crashes. Those 78 are crashes the agency believes the carrier could have prevented. That's a much sharper signal than the raw count would suggest.

When a plaintiff's lawyer sits across from you in deposition and asks "did you review the carrier's preventable-crash history before tendering the load?" — the right answer is "yes, here's the file." Not "I saw the crash count was high so I asked the carrier about it."

A concrete scenario

You're vetting MC-1392418 / DOT-3017294. Crash file shows three crashes in the last 24 months. SAFER snapshot shows Crash Indicator at 71% (above the 65% intervention threshold for general carriers).

The number alone says: stay away. But pull the actual crash detail from SAFER's Crash Report — or from the FMCSA crash file dataset on the DOT Open Data Portal — and you might see:

• Crash 1, 14 months ago: T-boned by a delivery van that ran a red light. Eligible CPDP type; carrier filed DataQ; designation **Non-Preventable**.
• Crash 2, 9 months ago: minor parking-lot scrape with another tractor at a truck stop. Driver was stationary. **Non-Preventable**.
• Crash 3, 4 months ago: driver lost-control on icy I-80, no other vehicle involved, no injuries. **Preventable** — the carrier didn't contest, conditions were known.

So the "real" preventable-crash count for this carrier is **1** out of 3. That's a meaningfully different picture than what the SAFER summary number suggested. The Crash Indicator percentile of 71 is computed before all the non-preventables are stripped (some take months to be reviewed). The percentile may drop materially after the DataQ reviews finalize.

Or the picture might go the other way. Same carrier, same three crashes, all three "Preventable" because the carrier never filed for review and the surface circumstances suggest driver fault. The 71% Crash Indicator just turned into a much sharper warning.

The crash COUNT in isolation gave you no way to tell which scenario you're looking at. The crash FILE — with preventability designations — is the actual decision-support data.

How to actually pull this

Three options, depending on volume:

For ad-hoc lookups, the FMCSA SAFER Company Snapshot at safer.fmcsa.dot.gov shows individual crashes with their CPDP designation in the Crash Report section. You can pull this in about a minute per carrier.

For programmatic vetting, the FMCSA crash data file is published on data.transportation.gov and includes the preventability field. If you screen a lot of carriers, you join the broker's MC# list to the crash file by DOT and aggregate preventable vs non-preventable per carrier. This is the kind of layer DOTScreener builds.

For ongoing monitoring, set up a quarterly DataQ review reminder for any carrier you book repeatedly. Many carriers have crashes that COULD have been filed Non-Preventable but never were — sometimes because the small carriers just don't know about the program, sometimes because they didn't have a safety director to file. If you're tendering loads to a carrier whose record has crashes that look eligible, prompting them to file a DataQ is in both of your interests.

The regulation, in plain English

The Crash Preventability Determination Program isn't an FMCSR section — it's a program FMCSA runs through the DataQs system at dataqs.fmcsa.dot.gov. The underlying authority is the agency's general crash-data-quality mandate under 49 U.S.C. § 31106 (motor carrier safety data) and the SAFER / CSA framework.

What this means at load-tender time: every crash on a carrier's public record has a preventability classification — Preventable, Non-Preventable, or Undecided. The Crash Indicator BASIC percentile is computed after non-preventables are removed. A defensible diligence file documents the actual breakdown of preventable vs non-preventable crashes, not just the total.

How I document this

For any carrier with crashes in the 24-month window:

1. **A note of the total crash count** plus the breakdown by preventability designation (Preventable / Non-Preventable / Undecided).

2. **For carriers with 2+ Preventable crashes** in the window: an explicit rationale paragraph for why I'm tendering anyway — what mitigations I'm requiring (enhanced attestation, restricted equipment, no high-value cargo on the first load, supervisor approval at our end).

3. **For carriers with 1+ Non-Preventable crashes that haven't been DataQ'd** but look eligible: a note that I flagged it to the carrier with a recommendation to file. (It improves both of our positions.)

4. **The Crash Indicator BASIC percentile** at the time of the lookup, with a note that "this percentile is computed post-non-preventable-removal" so the file shows I understood what I was reading.

The brokers who skip preventability analysis are the brokers whose deposition prep takes three days instead of one. The carrier file is the case, and the crash file is the part of it that gets the most attention from plaintiffs.

— Mason Lavallet

Founder, DOTScreener.com

---

Sources

• [FMCSA — Crash Preventability Determination Program (CPDP)](https://www.fmcsa.dot.gov/regulations/csa/crash-preventability-determination-program)
• [FMCSA DataQs — Request for Data Review](https://dataqs.fmcsa.dot.gov/)
• [49 U.S.C. § 31106 — Motor Carrier Safety Data](https://www.law.cornell.edu/uscode/text/49/31106)
• [FMCSA SAFER Company Snapshot](https://safer.fmcsa.dot.gov/CompanySnapshot.aspx)
• [FMCSA Crash File on data.transportation.gov](https://data.transportation.gov/Trucking-and-Motorcoaches/Crash-File/aayw-vxb3) — the dataset includes the preventability field
• [Hanson Bridgett — Montgomery v. Caribe Transport II analysis](https://www.hansonbridgett.com/publication/260514_8509_supreme-court-faaaa)