Truck accident and catastrophic injury statistics

01 How to use this section

If you came here looking for a figure to cite, the honest advice is to take it from the page that carries it rather than from a summary, and to take the year with it.

If you came here because you found a startling number somewhere else and want to know whether to believe it, start with the next section. Most of the striking claims in this field trace back to a single federal study, and most of them misstate what it found.

02 The most misquoted number in trucking

This matters for a practical reason. Figures presented as cause findings get quoted in demand letters and repeated on hundreds of websites, and the moment somebody checks the source the argument weakens rather than strengthens.

There is an awkward footnote to this, and we would rather set it out than tidy it away. The agency's own press release announcing the study was headed with the statement that driver behavior causes most truck crashes. So the framing the report cautions against appears in the publisher's own publicity. Both are the agency's words. Anyone quoting a percentage from this study should know that the report and the press release do not say the same thing, and should be ready for the other side to point that out.

03 Where the federal numbers come from

Knowing which one a figure came from usually tells you what it can support, and most published truck accident statistics come from just two of them.

• The fatal crash census. Operated by the highway safety agency, it aims to record every fatal crash on public roads. This is the most reliable truck accident data available and it is why death counts are stated confidently.
• The annual crash facts compilation. The motor carrier agency publishes a yearly volume drawing together crash, injury and exposure figures for large trucks and buses. It is the usual origin of headline national numbers.
• Carrier inspection and violation records. Roadside inspections, violations and out-of-service results are recorded against individual operators and can be searched by company. This is the only source on the list that speaks to one defendant.
• Causation research. Occasional, expensive, sample-based studies of why crashes happen, of which the study quoted above is the best known. Rich, dated, and easy to misread.

Injury figures are worth a separate word. Deaths are counted; injuries are largely estimated from samples of police reports, and the definitions of what counts as an injury are not uniform. Any set of truck accident statistics that presents injury totals to the same precision as death totals is overstating what the underlying data supports.

04 What the data cannot tell you

This is the limitation that matters most to somebody reading these pages after a crash.

• Fault is a legal conclusion. Crash datasets record circumstances, not findings of liability, and the two are decided by different processes.
• Settlement values are not in any federal dataset. Anybody publishing an average settlement is not drawing it from public crash data.
• Under-reporting is real. Crash counts depend on police reports being made and coded, and less severe collisions are least reliably captured.
• Lag is unavoidable. National figures are published well after the year they describe, so the most recent complete year is never the current one.

One historical figure is worth quoting for scale, expressly as history rather than as the present position. A 2009 report to Congress recorded that in each year of the preceding decade over 4,800 people have been killed and over 100,000 people injured in crashes involving large trucks. That is a statement about the late 1990s and 2000s. It is not a current count, and we have not used it as one.

05 The research pages

06 The reference pages

07 How to check a figure before you use it

Five steps, and they apply to figures on this site as much as to figures anywhere else.

08 How we source these pages

The standard exists because this field is full of figures that have been copied so many times their origin is untraceable.

• We quote, and we link. If a source says something useful, its words appear here rather than our summary of them.
• We date every figure. Because a number without a year is a number that will be wrong eventually and quoted anyway.
• We name what we did not check. Where we have not read a provision or verified a status, the page says so rather than implying completeness.
• We do not publish averages we cannot source. Particularly settlement averages, which do not exist in public crash data at all.

09 Red flags in statistics you will see elsewhere

The first is so common that it has become the ordinary way this field talks, which does not make it right.

A closing note on the purpose of all this. Nothing in a national dataset will tell you what your own case is worth, and a page of truck accident statistics is not a substitute for someone looking at your facts. What the numbers are good for is orientation, and for recognising when an argument being made to you rests on a figure that does not say what the person quoting it thinks it says. If you want someone to look at your actual facts rather than a national average, the button below opens a short private form and costs nothing.

Sources and authorities

Government publications, quoted and linked. Federal reports are public information.

Why this page carries no current national totals. Figures go out of date and then get quoted anyway. Earlier in the building of this site we published a figure that had been superseded four weeks before the page went live, which is the kind of error that is easy to make and hard to notice. National counts therefore live on the child pages listed above, where they can be updated as each agency publishes, and this page confines itself to explaining what those numbers are and are not.

On the tension in the causation study. We have quoted both the report's caution and the press release that ignores it. That is not an attempt to score a point against the agency: the study is careful, extensive and genuinely useful, and the caution is exactly the kind of thing a serious researcher writes. The problem is downstream, in the thousands of pages that quote the percentages without the caveat. Anyone using a figure from that study in a real dispute should read the methodology first.

What is not here. We have not summarized the four data families in numeric detail, because that belongs on the pages that carry the numbers. We have published no settlement averages, because no public crash dataset contains settlement values. And we have not analyzed any dataset ourselves; everything above describes what published sources say about their own methods.

Common questions, answered

General information, not legal advice. National data does not establish anything about an individual case, so check anything here against your own state and your own attorney.