And Then You're Dead

The Autopsy / Case File · Accidents & Injury

The Breed on the Certificate

Every fatal dog attack ends with a coroner's report, an animal control incident file, and, almost always, a news story — and each of those three documents can list a different breed for the same dog. Only one of them is usually checked against anything more rigorous than an eyewitness's best guess or a photograph run through the assumptions of whoever's looking at it.

A landmark CDC study of 238 fatal dog attacks from 1979 through 1998 found pit bull-type dogs and Rottweilers involved in 51 percent of deaths — a finding cited constantly in the decades since, attached to a set of caveats from the CDC itself that get cited far less often. A follow-up review of 256 fatal attacks from 2000 through 2009, published in the Journal of the American Veterinary Medical Association, found that 90 percent of the dogs involved had been confidently assigned a single breed by media coverage, while documented proof of that breed — pedigree records or DNA testing — existed for only a small fraction of cases. And when the same researchers identified the seven circumstantial factors that actually co-occurred in the overwhelming majority of fatal attacks, breed wasn't one of them.

And Then You're Dead · September 2026 · 14 min read

Join the discussion →
A beagle on a leash standing alert on a paved path, looking off to the side
Breed identification research doesn't single out any one type of dog — it's about how unreliable the label itself turns out to be, for any dog, once researchers check it against something more rigorous than a witness's description or a news photo. Photo: Daniel Flathagen, CC BY-SA 2.0, via Wikimedia Commons.

Fatal dog attacks are rare relative to the roughly 90 million dogs living in American households, but each one generates intense public interest in one specific question: what breed was the dog. That question has driven three decades of research, several hundred municipal breed-specific ordinances, and a federal agency's own retreat from collecting the data that started the whole conversation — a retreat worth understanding in detail, because it turns on a methodological problem that has only gotten more, not less, documented since.

The Study That Started It

In 2000, CDC researchers published a review of dog-bite-related fatalities recorded from 1979 through 1998, identifying 238 deaths involving at least 25 different breeds. Pit bull-type dogs and Rottweilers accounted for more than half of those deaths, 122 of 238, or 51 percent — the single statistic that has anchored breed-specific legislation debates ever since. But the same paper, in its own discussion section, laid out exactly why that number should be handled carefully: breed identification, the authors wrote, depended on witness and media reports, which are unreliable, and a raw death count by breed can't account for how many dogs of each breed existed in the country to begin with, meaning a breed with a large population will naturally accumulate more incidents than a rare one even if neither is actually more dangerous per capita. The researchers added a more specific caution about their own method: to the extent that attacks by one breed were more newsworthy than attacks by another, their approach — built substantially on searching news accounts — may have produced a systematically skewed count by breed, over-representing whichever breeds journalists and editors were already primed to report on. CDC stopped collecting fatality data broken out by breed after this study, a policy position it has held for a quarter century since.

What "Breed" Actually Meant in the Case Files

A 2013 follow-up study in the Journal of the American Veterinary Medical Association, examining 256 fatal dog attacks from 2000 through 2009, went further than the 2000 CDC study by checking media-reported breed against documentary evidence. The mismatch was substantial. Ninety percent of the dogs involved in these cases had been characterized in at least one media report using a single breed descriptor — the kind of confident, purebred-sounding label ("a pit bull," "a Rottweiler") that implies certainty. Population data from the same period, however, indicated that roughly 46 percent of dogs in the United States were mixed breed, not purebred — meaning either media breed-labeling was being applied far more loosely than the confident single-breed language suggested, or purebred dogs were wildly overrepresented among fatal attacks specifically, a possibility the researchers considered unlikely given the photographic evidence available to them in most cases. Actual documented evidence of breed — pedigree papers, breeder records, or DNA testing, rather than a witness's or reporter's visual impression — existed for a small minority of the 256 cases: 19 dogs in single-dog attacks and 28 dogs across multiple-dog attacks had that kind of documentation, out of several hundred dogs involved overall.

0
of dogs in a 256-case fatal-attack review confidently assigned a single breed by media reports
0
of the same cases showing at least 4 of 7 identified co-occurring preventable factors — none of which was breed

The Seven Factors That Were

Rather than treating breed as the organizing variable, the 2013 study identified seven specific, verifiable co-occurring circumstances present across the case files: no able-bodied adult present to intervene; no familiar relationship between the victim and the dog; the victim unable to interact appropriately with the dog, whether due to age under 5 or a compromised physical or cognitive capacity; the dog kept isolated from routine positive human interaction as a "resident dog" rather than integrated into the family as a companion animal; the dog not spayed or neutered; documented evidence of prior owner mismanagement, such as a history of the dog running loose or previous bite incidents; and a documented history of abuse or neglect by the owner. At least four of those seven factors were present together in 80.5 percent of the 256 cases studied — 206 out of 256. Individually, the factors were common: 87.5 percent of the dogs were male, 84.4 percent were sexually intact, 76.2 percent were kept as resident rather than family dogs, 37.5 percent of cases showed documented owner mismanagement, and 21.1 percent involved a documented history of abuse or neglect. Breed was not among the seven factors the researchers identified as co-occurring with fatal outcomes.

Four or more of seven specific, verifiable circumstances were present in 80.5 percent of fatal dog attacks. Breed wasn't one of the seven.

Why Even Trained Eyes Get It Wrong

The gap between confident media breed labels and documented breed evidence isn't unique to fatal-attack case files — it shows up in controlled research settings where the comparison can be made directly against DNA testing. A study led by a University of Florida shelter medicine researcher compared breed assessments made by 16 shelter staff members, including four veterinarians across four shelters, against DNA profiles for the same 120 dogs. For dogs with confirmed pit bull-type heritage in their DNA, staff correctly identified that heritage visually only 33 to 75 percent of the time, depending on which individual staff member was doing the assessing — and different staff evaluating the exact same dog at the exact same shelter often reached only moderate agreement with each other. A separate, larger study comparing shelter breed assignments against genetic testing found staff could correctly match at least one breed present in a dog's genetic heritage 67.7 percent of the time, but that figure collapsed to 10.4 percent when staff were asked to identify more than one breed contributing to a mixed-breed dog's ancestry — the more common case, given the population-level mixed-breed rate cited above. If trained shelter professionals working under controlled conditions, with the dog physically in front of them, disagree with DNA results and with each other at these rates, a news reporter working from a witness description or a single photograph after a fatal attack is working with meaningfully less reliable information still.

What the Agencies Actually Recommend

CDC's position, unchanged since it stopped publishing breed-specific fatality counts after the 2000 study, is that its own historical data should not be used to justify breed-specific policy, because the data doesn't demonstrate that any breed is inherently more dangerous than another once ownership and management factors are accounted for. The American Veterinary Medical Association holds the same position for the same underlying reason: its published guidance states plainly that it isn't a dog's breed that determines whether it will bite, but the dog's individual history, socialization, and management — precisely the seven factors the 2013 study found actually clustered around fatal outcomes. Neither position claims breed is entirely irrelevant to a dog's physical capacity to cause serious injury once a bite occurs; a larger, stronger dog can do more damage per bite than a smaller one, regardless of temperament. What both agencies have concluded, based on the same body of evidence reviewed here, is that breed identification itself is too unreliable, and breed alone too weak a predictor next to ownership and management factors, to serve as the basis for the kind of policy — breed-specific bans and restrictions — that continues to be proposed in its name.

And Then You're Dead

Every fatal dog attack produces a name for the breed involved within hours, usually from someone who saw the dog once, under the worst possible circumstances, and had no DNA kit in hand. The seven factors researchers actually found repeating across hundreds of case files — an unsupervised dog, an unfamiliar victim, a history already flagged before the fatal day — took years of case-by-case documentation to establish. The breed took one news cycle, and stuck regardless.

That's it. That's the whole thing.

Comments

Loading…

Comments are anonymous and moderated.

Sources