The Citation Is the Warning

Federal inspection records show crash risk spikes in the weeks right after a citation — and that the warning falls hardest on the carriers who can least afford it. We pulled ten million of the government's own roadside records to find out how strong that signal really is.

Every year, federal inspectors pull over trucks on the shoulder and write up whatever they find wrong — a burned-out taillight, a driver who's pushed past the legal hours limit, a brake that's worn thin. Most of the time, that citation gets filed and forgotten. But when we lined up 7,658,563 of those citations against what happened to the same carriers afterward, a pattern showed up that doesn't fade: the citation is where the story starts, not where it ends.

1. Crash risk is highest right after a citation, and fades over time. We tested this across 22 different violation types, from worn brakes to falsified driving logs, and the shape of the curve holds every time — the days immediately following a citation carry the most risk, then the risk tapers off the further out you look.
2. Small trucking companies are cited — and pulled off the road — far more often than large fleets, even after controlling for the fact that we're comparing citations per inspection, not just raw totals. This holds across every violation category we checked, not only hours-of-service.
3. We can't directly test whether large carriers simply get more citations dismissed than small ones do — the government doesn't publish that data. But the evidence we can see points toward a real difference in how the trucks are run day to day, not large companies being better at disputing tickets. We walk through why below.

Three terms, before you read on

For each of the 6 violation categories, we took every citation on record and checked whether that carrier had a crash in the following six months. Two things are deliberately excluded, because both would distort the picture: inspections that only happened because of a crash (that would make "crashes follow inspections" a circular finding), and any crash on the same day as the citation itself (too close in time to be a meaningful early-warning signal — more on both in "How we built this," below). Of the 7,658,563 citations, 851,598 were for hours-of-service alone.

A closer look: hours-of-service, day by day

The chart below shows, for hours-of-service citations specifically, how many crashes happened on each day following the citation, colored by severity. The tall bars on the left tell the story: risk is concentrated in the days immediately after.

By federal rule, a crash only gets officially recorded at all if it involved a fatality, an injury, or a vehicle towed away — a minor fender-bender with neither doesn't appear in this data. That's a limit of what's publicly trackable, not something we chose.

"Hours-of-service" isn't one thing — it breaks down into a handful of distinct situations, from driving while fatigued to a logbook that simply wasn't filled out correctly. Driving while fatigued and filing a false driving log carry the strongest link to a later crash; logbook paperwork errors carry the weakest.

Share followed by a crash within six months, ranked

Vehicle-condition citations are the single largest category by volume. Split by the type of equipment involved, brakes and tires carry a meaningfully stronger link to a later crash than paperwork-adjacent issues like inspection documentation.

Share followed by a crash within six months, ranked

Sometimes an inspector writes up more than one kind of problem on the same truck in the same stop — worn tires and an hours-of-service violation, say. Does getting flagged for both at once predict more risk than either alone, or is that just two coincidences sharing a page? We checked at three follow-up windows, so no reader is stuck trusting a single cutoff.

How much extra risk does the combination add?

Each cell is how many percentage points higher the "both at once" crash rate is than whichever single problem was already riskier on its own. Darker red means the combination adds more risk than either problem alone; pale means it adds little or nothing.

Yes — clearly, and across every violation category, not only hours-of-service. A single-truck owner-operator and a 500-truck fleet don't get inspected the same number of times, so counting total citations would just measure who's on the road more. We measure citations per inspection instead, so both are held to the same yardstick.

We also checked what happens to a company's presence in the inspection system after an hours-of-service citation, compared to a company that had a routine, citation-free inspection. Rather than a crash, the outcome here is simpler: does this company keep showing up in federal inspection records at all over the following year, or does it go quiet?

This isn't proof a company formally closed — going quiet in federal inspection records could also mean a change of ownership, a change of authority, or simply less time on the road. But it's a meaningful signal in its own right: a citation is followed not just by elevated crash risk in the months after, but by a real chance the company doesn't continue as an active carrier at all. The pattern holds even looking at single-truck operators specifically.

The federal government lets carriers formally dispute a citation after the fact, through a system called DataQs. If a dispute succeeds, the record can be corrected or removed. In principle, if large carriers use this more often or more successfully, that alone could explain part of the gap above — without any real difference in driving behavior.

We can't test that directly, because the government doesn't publish which disputes were filed or how they turned out — that's visible only to the company that filed and the agency reviewing it. The pulled-off-the-road comparison above is the closest available evidence: it's a real-time, on-the-spot outcome rather than a later paperwork judgment, and the gap there is even larger than for ordinary citations — which points toward a real behavioral difference.

We don't count inspections that only happened because of a crash
If a truck is inspected right after a crash, inspectors dig much harder and find far more problems — in our data, that pushes some violation types to be found roughly 40 times more often than during a routine, unprompted inspection. Counting those would make our finding partly a restatement of "crashes get investigated," rather than "citations predict crashes." We remove all crash-triggered inspections before doing any counting.

We don't count a citation and a crash on the very same day as connected
A citation written up the same day as a crash is usually paperwork from that same incident, not an early warning of something that happened later. We only count a crash as "following" a citation if it happened on a later day.

Multiple citations on the same stop, for the same kind of problem, count once
If an inspector writes up five separate lighting defects on one truck during one stop, we count that as one "lighting problem" event that day, not five.

Minor, no-injury, no-tow crashes aren't in this data at all
By federal rule, a crash only becomes an official, trackable record if it involved a fatality, an injury, or a vehicle towed away. Every "crashed" figure in this report already implies at least one of those three — a minor fender-bender with neither simply isn't something we, or FMCSA, can see.

The comparison window is six months, and the small-carrier comparison uses a fixed 2-year period
We look for a crash in the six months following each citation. For the small-vs-large comparison specifically, both the citation count and the inspection count are measured over the exact same fixed two-year window (mid-September 2024 to mid-September 2026), so it's an apples-to-apples comparison rather than two numbers measured over different periods.Why you can trust these numbers

Every number in this report was calculated directly from FMCSA's own inspection and crash records as of September 22, 2026.

• The detailed breakdown (22 specific violation types) and the high-level 6-category breakdown were calculated two separate, independent ways from the same underlying records — and every number they share in common came out identical between the two methods.
• We confirmed, across all 329,167 crash records checked, that every single one had at least a fatality, an injury, or a tow-away — exactly as the federal reporting rule requires. Zero exceptions, which is exactly what should happen if the data is being read and matched correctly.
• The logic used to sort citations into categories is the same logic that was separately tested against a large, independent set of real carrier records and matched perfectly.