The Crash Count on SAFER Is the Wrong Number to Look At

Two carriers on your load board for the same dry-van lane, next-day, 480 miles. Carrier A has 4 crashes in the last 24 months. Carrier B has 2. On a quick look, Carrier B wins.

But Carrier A's four crashes are all coded "tow-away" — vehicles disabled, no injuries, the kind of thing that happens backing into loading docks or clipping a mirror in a fuel island. Carrier B's two crashes include one coded as an injury crash at highway speed, 90 days ago.

Which carrier do you actually want on your freight?

The data to answer that question is sitting in SAFER, in a table most brokers skip entirely. The crash count is there. The severity breakdown — fatal, injury, tow-only — is right below it. The habit of reading the number and ignoring the breakdown is one of the most common vetting gaps I see, and it's also one of the cleanest paper-trail failures once a case gets to discovery.

What SAFER actually shows you

Pull any carrier's FMCSA SAFER Company Snapshot and scroll to the crash section. There's a 24-month rolling window. Total crash count at the top. Underneath: three rows broken out as Fatal, Injury, and Tow-Away.

Every crash in SAFER crosses the DOT-reportable threshold under 49 CFR § 390.5 — meaning it involved either a fatality, a bodily injury requiring medical treatment away from the scene, or disabling damage that required the vehicle to be towed. So "tow-away with no injury" is the floor of what gets counted. Fatal and injury crashes are explicitly coded as distinct rows in the snapshot table.

Most brokers read across the total and move on. The total is the least useful number in that table.

The definition tells you why the breakdown matters

49 CFR § 390.5's definition of "accident" is the key. Three qualifying scenarios: a fatality, an injury requiring off-scene medical treatment, or disabling damage requiring tow. These are not equivalent events. A slow-speed backing incident that pushes a dock seal off-center and disables the trailer until a wrecker arrives is an "accident" under this definition. So is a highway-speed collision that kills a passenger car occupant. Both count once in the SAFER total.

So when someone tells you a carrier has 3 crashes, that sentence could describe any combination of those scenarios. Three tow-aways with no injury is fundamentally different from one fatality and two tow-aways. Three injury crashes with no fatalities is a different profile than six tow-aways with zero injuries.

At load-tender time, "how many crashes" is the wrong question. "What kind" is the question.

Why the Crash Indicator BASIC doesn't close this gap

The Crash Indicator BASIC percentile is useful — it scores a carrier against peers with similar exposure, adjusting for time-weighting and mileage. A carrier sitting in the 75th percentile has more crash exposure per unit of travel than 75% of comparable carriers, and that's worth flagging. But the percentile doesn't distinguish by severity. A carrier with six minor dock incidents can sit higher in Crash Indicator than a carrier with one catastrophic highway crash, depending on the peer group and timing.

There's also the sample-size problem. A carrier with low inspection counts may show "not shown" in the BASIC — too few data points to generate a reliable percentile — but still have an injury crash coded in the snapshot. The BASIC is absent. The crash is not. If you're only looking at the BASIC alert, you miss it.

The severity breakdown is what gives you context the BASIC can't supply. If Crash Indicator is elevated, the breakdown tells you whether that's because of repeated low-severity incidents or a more serious event history. If Crash Indicator is clean — or absent — the breakdown can still surface a high-severity crash that didn't move the BASIC enough to trigger an alert.

What happens when you skip it in discovery

A carrier — call them MC-1491832 / DOT-3562107, two years of authority, small fleet — has two crashes in SAFER. A broker pulls the snapshot, notes "2 crashes" in the carrier file, and tenders a load. Six weeks later, that carrier is involved in a serious accident. A wrongful-death case develops with a demand north of $2.5 million.

Discovery turns up the broker's file. The plaintiff's attorney pulls the SAFER snapshot as it existed on the date of tender. One of those two crashes from the prior history was coded as an "injury" crash on a state highway 11 months before tender. The other was a tow-away at a terminal. The file says "crash history reviewed — 2 total." No breakdown. No note on what the injury crash involved. No indication the broker understood there was a distinction.

In deposition: "I checked and it said 2 crashes." That's a losing answer when the plaintiff is arguing a competent vetting process would have noticed an injury crash in recent history and either dug deeper or documented a rationale for proceeding.

The injury crash alone isn't necessarily disqualifying — a carrier with one prior injury crash, clean BASIC scores, good inspection history, and 24 months of continuous authority might still be the right call on most lanes. But the broker who never noted the severity breakdown can't show they understood what they were looking at. The documentation gap is where these cases often turn.

Reading the breakdown in practice

When you're in the SAFER snapshot, the crash table breaks down into Fatal / Injury / Tow-Away. Here's how to use those three rows:

If Fatal is nonzero, that's an immediate deeper-look situation. What happened, how long ago, what has the carrier done since? A fatal crash 23 months ago followed by 23 months of clean inspections and a fully clean Unsafe Driving BASIC is a different situation than a fatal crash 4 months ago. Document what you find.

If Injury is nonzero, it doesn't automatically disqualify the carrier — but it means you should look harder. Was it recent? Is the Crash Indicator BASIC elevated alongside it? What does the carrier's inspection history look like in the months after? A phone call to the carrier to understand what happened is appropriate on any load worth more than a few dollars in liability exposure.

If all crashes are tow-away with zeros in Fatal and Injury, the picture is meaningfully less alarming. You still note the breakdown. But the story you're telling in the file is different: these events crossed the DOT-reportable threshold but didn't rise to injury level.

Also factor in size and recency. A carrier running 30 power units with 4 tow-aways in 24 months has different exposure math than a carrier with 3 trucks and 4 crashes. SAFER doesn't show you when in the 24-month window the crashes occurred — the window is flat — so if the count is elevated, dig into the inspection data to get a sense of timing. A cluster of crashes 20+ months ago that stopped after a safety management change is a different conversation than a cluster in the last six months.

How I document this

In every carrier file, the crash entry looks like this:

SAFER crash history (pulled [date]): Total: [X]. Fatal: [X]. Injury: [X]. Tow-away: [X]. [If Fatal or Injury is nonzero: what the situation appeared to be, whether a call was made, any RDR preventability determination visible.]

That's the whole thing. Three extra fields and thirty seconds. It turns "crashes reviewed, 2 total" — which shows nothing about understanding — into a record that proves you looked at the breakdown, not just the count.

If there's no crash history at all, the entry still belongs in the file: "SAFER crash history (pulled [date]): Total: 0." I've written before about why zero isn't a free pass, but it's still a datum that should be timestamped in your file.

The reason this matters: a subpoena produces the carrier file. The plaintiff reconstructs what you looked at and what you missed. A file that shows "0 fatal, 1 injury, 1 tow-away — spoke with safety manager, confirmed injury crash was a rear-end at a red light (counterparty at fault), no preventability determination filed" is defensible. A file that shows "2 crashes" with nothing else tells the jury you either didn't look carefully or didn't think it mattered. After Montgomery v. Caribe Transport II, that's not a posture you want to be in.

The number most brokers skip is right there

It's not buried. It's not a separate FMCSA data product you have to request. It's three rows below the crash total in the same table you're already reading. Fatal. Injury. Tow-away. Those three rows turn a count into a signal.

Read the breakdown.

— Mason Lavallet

Founder, DOTScreener.com