Nobody knows how many hawks were killed by cars in the United States last year. Nobody knows the number for owls, foxes, or bears either. No agency keeps that count, and anyone who hands you a clean national figure for those four animals is guessing.
Start there, because the gap is the story. A wildlife-vehicle collision — WVC, in transportation language — gets recorded when it damages a vehicle, injures a person, or generates a claim. The system was built to track crashes, not deaths. A bear bends a bumper and enters a database. A barn owl does not.
why there is no national body count
American roadkill data comes from three uneven streams: police crash reports, insurance claims, and carcass surveys run by agencies or volunteers. Each sees a different slice, and none of them was designed to count a species.
The UC Davis Road Ecology Center says it plainly about two of the most charismatic animals in California: there is no formal program, system, or requirement to report mountain lions and black bears killed on roads. What exists are incidental reports. "As such," the 2023 report notes, "we only know the minimum killed each year."
Even the best global compilation is a dataset, not a total. Global Roadkill Data, published in Scientific Data, assembled 208,570 georeferenced records of dead terrestrial vertebrates from 54 countries between 1971 and 2024. That is the largest roadkill dataset in existence. It is also, by construction, a record of what someone happened to find and log.
So the honest framing is this: for large mammals we have crash economics, for birds we have statistical extrapolations, and for hawks, owls, foxes, and bears we have mechanisms, state tallies, and single-highway studies. That is enough to act on. It is not a census, and calling it one would be the easiest way to lose a biologist's trust in the first paragraph.
what we can actually count
Here is the full picture of what is known, with the caveat attached to each number rather than buried in a footnote.
| what gets counted | best available number | what it is not |
|---|---|---|
| Large-animal collisions, US | 1–2 million per year | Reported and estimated crashes, not total roadkill |
| Vertebrates killed on US roads | ~365 million per year | An extrapolation carried in the FHWA report, not a count |
| Birds killed by vehicles, US | 89–340 million per year; separately, 96 million (95% CI 59–200) | Two methods, two answers, both wide |
| Owls, US | No national total | Best evidence is British and single-highway |
| Hawks, US | No national total | The mechanism is documented; the count is not |
| Black bears, Florida | 289 road deaths in 2025; 4,694 over 2006–2025 | One state's known deaths |
| Black bears and mountain lions, California | ≥1,024 bears and 470 lions, 2016–2022 | A reported minimum, not a count |
| Foxes, US | No national total | Study-level percentages, not totals |
how many animals are killed by cars in the us?
Between one and two million wildlife-vehicle collisions involving large animals occur annually in the United States, and the number is rising. That is the Federal Highway Administration's finding in its Wildlife-Vehicle Collision Reduction Study, the report to Congress that remains the reference point for American road ecology.
Those collisions cost people roughly 200 lives, about 26,000 injuries, and a best-estimate total of $8.388 billion a year. More recent FHWA program materials cite a higher annual cost; the study figure is the one with the published method attached, so it is the one used here.
For everything smaller than a deer, the same report carries an estimate of 365 million road-killed vertebrates a year in the United States. Birds are the best-studied slice of that. Loss and colleagues reviewed 13 studies and estimated 89 to 340 million birds killed annually on U.S. roads — the figure the U.S. Fish and Wildlife Service cites. A 2024 study in the Journal of Field Ornithology took a different route entirely, asking drivers to log collisions across 2.27 million kilometers of driving, and landed at 96 million birds per year with a 95% confidence interval of 59 to 200 million.
Two independent methods, overlapping ranges, no false precision. "Tens to hundreds of millions of birds" is the defensible sentence.
how many owls are killed by cars?
There is no national owl count. There is a well-documented reason owls die on roads more than their numbers suggest they should.
The U.S. Fish and Wildlife Service names three species specifically: Great Horned Owls, Barn Owls, and Short-eared Owls commonly forage at the same height as vehicles. Mowed verges grow the grass that feeds the small mammals that draw the owl to hunt at bumper level, in the dark, along a corridor of moving steel. Habitat management creates the hazard; the hazard is not an accident of bad luck.
Britain has studied one of those species long enough to produce numbers. The Barn Owl Trust estimates that Britain's roughly 4,000 pairs produce about 12,000 young in a typical year, and that 3,000 to 5,000 of those young are killed on roads. Major roads — motorways and dual carriageways — make up less than 2% of Britain's road network but account for over 90% of Barn Owl road casualties. Roughly half of all reported Barn Owl deaths are road casualties.
The United States has one comparable window. Along Interstate 84 in southern Idaho, researchers surveying road-killed wildlife found that American Barn Owls outnumbered every other identified vertebrate species on the highway. They made up more than 25% of all individuals recovered and 73.6% of the road-killed birds. That is one interstate over two survey periods, not a national rate — and it is a real, published, unhappy number for anyone who assumed owls were a rounding error.
Roads are not the only bill we hand this bird. Rodenticide moves up the same food chain and kills the owls doing pest control for free — a different invoice for the same species, covered in the owl we poison while paying for pest control.
how many bears are killed by vehicles?
No national number exists here either. Two states publish enough to be useful.
Florida keeps the most complete bear mortality record in the country. The Florida Fish and Wildlife Conservation Commission recorded 289 bears killed on roads in 2025. Across 2006 through 2025, roads killed 4,694 of the roughly 5,900 bears whose deaths the state knows about. Roads are also where carcasses get found, so documented deaths over-sample the highway. Even with that bias, vehicles are the leading documented cause of death for the Florida black bear. The statewide population is estimated at about 4,000 animals.
California shows the same problem with worse instrumentation. Between 2016 and 2022, at least 1,024 black bears and 470 mountain lions were reported killed on California roads, compiled from volunteer observations, highway patrol reports, wildlife agency records, and researchers. That is a floor, not a total. The Road Ecology Center is explicit that there is no way to know the actual number.
do hawks and foxes die on roads?
Yes, and neither has a national total.
Hawks die on roads for a reason that compounds every other death on this page: raptors, eagles, vultures, and corvids are drawn to roadside carcasses. A deer struck at midnight becomes the meal that pulls a red-tailed hawk into traffic at dawn. Roadside perch structures — poles, wires, signs — put the bird where the food is. The Fish and Wildlife Service documents the behavior. Nobody documents the count.
Foxes are the same story with one useful exception. The endangered San Joaquin kit fox has been studied closely enough to show the spread: across the studies compiled in the FHWA report, vehicle mortality accounted for between 2% and 65% of recorded deaths, depending on the population and the road. Two percent in one study of 60 deaths; 65% in another of 23 adult deaths. After predation, vehicles are likely the second most common cause of death for the species. Red foxes appear in nearly every regional roadkill list in North America and Europe without ever resolving into a national figure.
That range — 2% to 65% — is the most honest number in this post. It says the answer depends entirely on where the road is.
the invoice, and what it is not
The crash bill is real, itemized, and already being paid. Around 200 people die and 26,000 are injured. FHWA's 2008 estimate put the annual bill at roughly $8.4 billion — a deer-collision figure that mixes vehicle repair, medical costs, a fatality share, towing, carcass removal, and a $2,000 "monetary value of animal" line. It is not a property-damage total, and it is not the worth of a bear. Every one of those payers is already funding the failure.
But that is the human side of the invoice, and it is not the worth of the animal. A bear is not an avoided repair bill. A Barn Owl is not a line item in a highway maintenance budget. FHWA's own per-collision math already prices a deer at $2,000 — which is exactly the collapse this post refuses. Pricing exists so that capital can see what it is destroying — the number is a bridge to a decision, never a claim about what the animal is worth.
And the crash bill is only the half we can see. The other half never generates a claim at all:
- A herd that stops using a winter range because a four-lane sits between it and the grass.
- A population split into two smaller populations, each with less genetic room to survive a bad decade.
- A mountain lion range in Southern California where vehicle strikes are likely the first or second leading cause of death, producing exactly the kind of segmentation and inbreeding that ends a subpopulation quietly, without a single carcass on the shoulder.
The carcass is the receipt for a transaction that already happened. The transaction was severing the route.
count what we can, then fund the route
The measurement gap is fixable and worth fixing — a mandatory reporting standard for large-carnivore road deaths would cost almost nothing and end the phrase "we only know the minimum." But better accounting does not save an animal. It only tells you what you lost with more precision.
What stops the invoice from compounding is structure plus land: a crossing over or under the pavement, fencing that funnels animals to it, and protected habitat on both sides so the crossing leads somewhere. A crossing without land is a bridge to a subdivision. Land without a crossing is an island with a highway for a shoreline.
That is what ensurance funds. Instead of paying for the crash after it happens, a coin or a certificate routes capital to a named route — the corridor land and the crossing structure — before the collision. Think of it as buying the prevention that the crash bill is currently buying nothing at all.
If you want to see what holding a named natural asset looks like in practice, start with the natural assets binder. If you want the engineering, read what a wildlife crossing actually is. If you want the reason any of it matters beyond the crash ledger, read to protect wildlife, protect how they move.
One place-based version of this exact problem is already documented: the Caloosahatchee crossing, where a river, a highway, and a panther population meet.