The “Dangerous Breed” Myth: Why Certain Breeds Are Wrongly Considered Dangerous – A Scientific Look

In brief

“Fighting dog” (in German, Kampfhund) is a social label, not a scientific one. The evidence is nuanced—and that is exactly what makes it convincing: breed differences in behavior exist and are partly heritable; no serious researcher claims that all dogs are a blank slate.

What matters, though, is that breed is a weak predictor of an individual dog’s behavior—especially when it comes to aggression. For many behavioral traits, the variation within breeds is very large and overlaps heavily between breeds. Not every single trait shows this pattern to the same degree, but the traits most closely related to aggression do. This is exactly where breed lists reach their limits: they judge an individual by a characteristic that predicts little in the individual case. This article clears up myths—on both sides of the debate.

The term “fighting dog” is one of the most emotionally charged in the world of dogs. American Staffordshire Terriers, pit bulls, Rottweilers, and Bull Terriers are often portrayed in public perception as especially dangerous across the board. What does the research say?

1. What this is about

1.1 The more precise question

Not: Are certain breeds dangerous? But rather: How well can the behavior of a particular dog be predicted from its breed? That is the question breed lists have to be measured against—because that is precisely what they claim to do.

1.2 How to read the evidence

Much of the literature is based on owner surveys and retrospective case analyses, not on controlled experiments. That is not a weakness of the researchers but a property of the subject: you cannot produce bite incidents experimentally. Causal claims should therefore be treated with corresponding caution—in both directions.

1.3 What this article does not claim

It does not claim that breed is meaningless. And it does not claim that how a dog is kept and raised explains everything on its own. Both statements are wrong, and both are made regularly in this debate.

2. “Aggression” is not “dangerousness”

The two terms are often lumped together in the debate—wrongly.

Aggressive behavior is a normal, communicative behavior that occurs in all mammals. It has causes: fear or insecurity, pain or illness, resource guarding, frustration, being overwhelmed, or territorial behavior.

A dog usually becomes dangerous when several factors come together: lack of control, inadequate socialization, mistreatment, deliberate training for aggression—often combined with great physical strength and poor impulse control. Breed alone is not a reliable risk factor here.

3. How the myth came about

3.1 An image that changed

In her widely noted account Pit Bull: The Battle over an American Icon, journalist Bronwen Dickey (2016) traces how the reputation of these dogs shifted. In the early 20th century, dogs of today’s pit bull type had a far more mixed, in part mainstream image in the United States. According to Dickey, the picture of the “born killer” is a more recent construct made of questionable science, media sensationalism, and fears around illegal dogfighting that were in part racially charged.

Staying honest, however, also means this: the often-repeated counter-narrative that pit bulls used to be loving “nanny dogs” is a myth as well—Dickey explicitly shows that this label has no historical basis. Neither the “killer” nor the “nanny on four paws” is a factual category.

3.2 The blind spot: Who actually identifies the breed?

Reports on bite incidents name certain breeds strikingly often. The problem is more fundamental than it first appears: breed attribution is usually based on appearance alone—and appearance is unreliable.

Gunter and colleagues (2018) checked shelter staff’s visual breed assignments genetically in two animal shelters. When a single primary breed was given, the assessment was correct in about two-thirds of cases; when several breeds were named, agreement dropped to about ten percent (10.1371/journal.pone.0202633). So a “pit bull” in a headline often isn’t one at all—and that affects a data basis on which a considerable part of the breed list debate rests.

4. What the research says

4.1 Why bite statistics by breed are so hard to read

Before we get to individual studies, one caveat that applies to every statistic of this kind. Raw case counts by breed are prone to misinterpretation for four reasons:

• The denominator is missing. Without solid figures on how many dogs of a breed are kept in the first place, no risk can be calculated from incident counts. A common breed shows up more often—even with the same risk.
• Ownership effects. Breeds are not distributed randomly among owners. Who gets which dog, and how they keep, secure, and train it, is connected to the choice of breed—and affects the outcome.
• Reporting behavior. An incident involving a listed dog is more likely to be reported, documented, and covered than the same incident involving an inconspicuous dog.
• Regional differences. Data collection, the definition of a reportable incident, and breed assignment differ between countries and between states or regions.

These four points do not make bite statistics worthless—they just make them something other than what they are usually used for.

4.2 The CDC fatality study, read correctly

A widely cited CDC study (Sacks et al., 2000) analyzed 238 fatal dog attacks in the United States between 1979 and 1998. Pit bull–type dogs and Rottweilers were indeed involved in more than half of these deaths. This figure is still cited in support of breed lists today—usually without the limitations the authors themselves pointed out:

• A population denominator is missing: without reliable figures on how many dogs of each breed are kept at all, no breed-specific risk can be calculated.
• Breed was mostly determined by appearance or from media reports.
• Fatal attacks are extremely rare and a tiny fraction of all bite incidents.
• Many different breeds were involved.

The authors explicitly concluded that breed-specific bans are not an appropriate approach to bite prevention. Anyone who cites this study as evidence for breed lists turns its conclusion on its head.

4.3 The better data basis

Thirteen years later, Patronek and colleagues (2013) examined the same question with better sources: 256 dog bite–related fatalities in the United States between 2000 and 2009, analyzed not from media reports but from investigative records (10.2460/javma.243.12.1726).

Two results stand out. First, the clustering of preventable circumstances: no able-bodied person present to intervene (87.1 percent), no familiar relationship between victim and dog (85.2 percent), dogs that were not neutered (84.4 percent), the victim’s limited ability to interact appropriately with dogs (77.4 percent), dogs without regular positive human contact (76.2 percent). In four out of five deaths, four or more of these factors occurred together.

Second, the breed question: reliable breed identification was possible in only 45 of the 256 cases—17.6 percent. Where media reports and official records could be compared, they disagreed for roughly 30 to 40 percent of the dogs. The authors’ conclusion: most of these deaths were characterized by coinciding, preventable factors—and breed was not one of them.

4.4 Individual variation outweighs breed differences

Precision matters here, because the evidence is often oversimplified—by both sides. The largest study on the subject to date surveyed the owners of 18,385 dogs and sequenced the DNA of 2,155 animals (Morrill et al., 2022; 10.1126/science.abk0639).

The findings in detail:

• Most behavioral traits are heritable, with heritability above 25 percent. Anyone who says it’s all down to upbringing is wrong.
• Even so, breed explains only about 9 percent of the variation in behavior between individual dogs.
• And the most important point for this debate: the less heritable and the less breed-typical a trait is, the worse breed predicts it. When it comes to how easily a dog is provoked by frightening or unpleasant stimuli, breed is almost uninformative. For the willingness to orient toward people, it is somewhat more informative.

That is the decisive finding: the very trait that breed lists aim to predict is the one for which breed tells you the least.

4.5 Context matters more than breed

In a large British survey, Casey and colleagues (2014) examined risk factors for human-directed aggression (10.1016/j.applanim.2013.12.003). Two findings are central.

First, aggression is context-specific: different risk factors were relevant for aggression toward family members than for aggression toward unfamiliar people. A dog that reacts aggressively in one situation often does not in others.

Second, all the measured characteristics together—including breed group, sex, neuter status, and owner characteristics—explained only a small part of the differences between aggressive and non-aggressive dogs. What stood out was the use of punishment-based training methods: it was associated with aggression—an association, not proof of cause. The researchers conclude that it is inappropriate to judge an individual dog’s risk of aggression by characteristics such as breed group.

4.6 What breed lists actually do

Gunter and colleagues (2016) examined what a label does. Dogs labeled as “pit bulls” in an animal shelter waited about three times as long for a home as similar-looking dogs with a different label. When the shelter stopped putting breed labels on the kennels, length of stay and adoption numbers improved (10.1371/journal.pone.0146857).

That is a measurable harm. A measurable benefit—a demonstrable reduction in serious bite incidents due to breed lists—has not been shown.

That still doesn’t settle the matter. Whether breed lists make sense legally and politically is a societal question that research alone cannot answer: it can say how well a criterion predicts, not what level of protection a society wants and what effort it is willing to accept for it. What research can do is put the debate on a solid factual footing.

5. Science and the law are two different things

One point that often gets lost in this debate: the professional assessment and the law in force are two different things, and for you as an owner, when in doubt, the law is what counts.

Many countries and municipalities have some form of breed-specific legislation (BSL). Germany is one example: there, each federal state decides which breeds are listed, which requirements apply, and how a temperament test is conducted. The rules differ considerably, and moving across a state border can completely change the situation for the same dog. Wherever you live, the reliable source for what applies in your case is the responsible local authority.

This section is a general overview and does not replace legal advice. From a professional standpoint, one thing can be said: an individual behavioral assessment reflects the risk posed by a specific dog much better than its breed does.

6. The human role

6.1 Early socialization

The first weeks of life—roughly the third to the sixteenth week—are a sensitive period. Positive experiences with people, dogs, and environmental stimuli lay the foundation for balanced adult behavior; if they are missing, insecurity and fear can develop, regardless of breed (Serpell & Duffy, 2014). More in Puppy Socialization and Puppy Training: Why the Start Matters.

6.2 Aversive training methods

Herron and colleagues (2009) analyzed 140 questionnaires from owners who had made an appointment at a behavioral medicine referral service. According to the owners, a considerable proportion of dogs reacted aggressively to several confrontational methods: hitting or kicking in 43 percent, growling at the dog in 41 percent, the alpha roll in 31 percent (10.1016/j.applanim.2008.12.011). How far this goes: these were dogs that already had a behavior problem—the figures apply to that group. More in Learning Theory in Dog Training and Aversive Training Methods in Dogs: Neurological Effects, Stress Responses and Long-Term Welfare Risks.

6.3 Pain as an overlooked cause

A frequently overlooked cause of sudden aggression is pain: osteoarthritis, dental problems, or hidden injuries can make a dog react sensitively to touch. Sudden changes in behavior should therefore always include a veterinary checkup. More in Recognizing Pain in Dogs and Chronic Pain and Behavior: The Neurobiological Link Between Osteoarthritis and Aggression in Dogs.

7. The individual, not the breed

The studies cited here consistently suggest that breed explains only a small part of an individual dog’s behavior—individual predisposition, experience, and environment carry considerable weight alongside it. Epigenetics also fits the picture: experiences can change gene activity without changing the DNA sequence itself. A dog that grows up feeling safe can develop differently from one that experiences neglect—with the same genetic predisposition. More in Epigenetics in Dogs: How Experiences Affect Their Genetic Makeup.

8. What this means in practice

8.1 For owners of listed dogs

The research does not exempt you from any legal requirement, but it puts into perspective what really matters: not proving that your dog is not a listed breed, but how it behaves in specific situations and how reliably you can handle it. That is also the basis on which a temperament test works.

8.2 For all other owners

The factors that cluster in the fatality analyses are not tied to any particular breed: no capable person nearby, no familiar relationship between child and dog, dogs without regular human contact. These are questions of how dogs are kept and supervised, and they apply to every dog. What to do if the worst happens is covered in When a Dog Bites.

9. Why the myth does harm

Blanket classification of certain breeds as dangerous has real consequences: shelter dogs are harder to place, euthanasia rates are higher in some countries, even completely unremarkable dogs are stigmatized, and responsible owners face restrictions.

There is another harm that is mentioned less often: breed lists create a false sense of security. If you believe the risk lies in certain breeds, you pay less attention, with all other dogs, to the factors that actually count—supervision, socialization, interactions with children, living conditions. It is precisely these factors that cluster in the fatality analyses.

10. “Fighting dog”? Or simply a dog?

Terms like “fighting dog” or “listed dog” are emotionally charged and professionally untenable—although “listed dog” at least describes a legal category, not a biological one. Dogs from so-called fighting dog breeds can, like all dogs, be loving family dogs, work in dog sports, serve as therapy or visiting dogs, and live peacefully with other dogs. For a deeper look: Breed and Behavior: Genetic Predisposition versus Individual Variation in Dogs and Aggressive Behavior in Dogs.

11. Where the evidence comes from

The data are better than you might expect on such an emotional topic—and they have one clear weak spot. The largest behavioral study covers more than 18,000 dogs, the best fatality analysis 256 documented cases based on investigative records rather than newspaper reports, and the aggression survey several thousand owners.

The weak spot lies elsewhere: almost everything is based on owner reports or retrospective reconstruction. Controlled experiments are neither possible nor ethically justifiable here. And the sore point of the entire breed debate is breed identification itself—if reliable assignment was possible in only 17.6 percent of cases in the best-documented fatality study, that affects every statistic built on such assignments.

There is also a geographic imbalance: almost all of the literature comes from the United States and the United Kingdom. For Germany, for example, with its state-level rules, temperament tests, and owner competence requirements, there are hardly any comparable studies.

12. What the research does (not yet) show

• Breed is not meaningless. Most behavioral traits are more than 25 percent heritable. Saying “it’s all down to upbringing” is just as much an oversimplification as the opposite position.
• The point is prediction in the individual case. Breed explains about 9 percent of behavioral variation and is not suitable for assessing the risk of a particular dog.
• Dangerousness is not measured directly. Much is based on owner surveys and retrospective case analyses, not on controlled experiments.
• There is no proof that breed lists work—what is documented, by contrast, are disadvantages for labeled dogs when it comes to adoption.
• Not which approach reliably reduces bite incidents. The fatality analyses show which circumstances cluster; whether serious incidents can be measurably reduced through particular regulations remains open.
• No data on how the rules work in practice in Germany. Whether temperament tests, owner competence certificates, and leash requirements are effective there has not been studied.

13. Conclusion

The “fighting dog” myth is not based on research but on media narratives, social fears, and political decisions. A dog’s breed says little about its individual behavior—individual predisposition, upbringing, experiences, and environment play a major role alongside it.

That is no license for carelessness. A powerful dog can do more damage than a small one, and anyone who keeps such a dog bears more responsibility. But that responsibility depends on strength and handling, not on the breed name—and it applies to the Labrador just as much as to the Staffordshire Terrier.

Dogs are not scapegoats for social fears. They are individuals and deserve to be judged as such.

Key takeaways

• Across 18,385 dogs whose owners were surveyed, breed explained about 9 percent of behavioral variation; for the threshold at which a dog is provoked, of all things, it was almost uninformative (Morrill et al., 2022).
• Most behavioral traits are more than 25 percent heritable—so breed is not meaningless, it just says little about the individual case.
• In 256 fatal bite incidents, reliable breed identification was possible in only 17.6 percent; in four out of five cases, four or more preventable circumstances coincided (Patronek et al., 2013).
• Visual breed assignments in shelters were correct in about two-thirds of cases for a primary breed, but in only about ten percent when several breeds were named (Gunter et al., 2018).
• Shelter dogs labeled as “pit bulls” waited about three times as long for adoption as similar-looking dogs without that label (Gunter et al., 2016).
• Aggression is context-specific; the use of punishment-based training methods was associated with aggression (Casey et al., 2014).

References

Casey, R. A., Loftus, B., Bolster, C., Richards, G. J., & Blackwell, E. J. (2014). Human directed aggression in domestic dogs (Canis familiaris): occurrence in different contexts and risk factors. Applied Animal Behaviour Science, 152, 52–63. https://doi.org/10.1016/j.applanim.2013.12.003

Dickey, B. (2016). Pit Bull: The Battle over an American Icon. Alfred A. Knopf, New York. No DOI has been assigned to this book.

Gunter, L. M., Barber, R. T., & Wynne, C. D. L. (2016). What's in a name? Effect of breed perceptions and labeling on attractiveness, adoptions and length of stay for pit-bull-type dogs. PLOS ONE, 11(3), e0146857. https://doi.org/10.1371/journal.pone.0146857

Gunter, L. M., Barber, R. T., & Wynne, C. D. L. (2018). A canine identity crisis: genetic breed heritage testing of shelter dogs. PLOS ONE, 13(8), e0202633. https://doi.org/10.1371/journal.pone.0202633

Herron, M. E., Shofer, F. S., & Reisner, I. R. (2009). Survey of the use and outcome of confrontational and non-confrontational training methods in client-owned dogs showing undesired behaviors. Applied Animal Behaviour Science, 117(1–2), 47–54. https://doi.org/10.1016/j.applanim.2008.12.011

Morrill, K., Hekman, J., Li, X., McClure, J., Logan, B., Goodman, L., … Karlsson, E. K. (2022). Ancestry-inclusive dog genomics challenges popular breed stereotypes. Science, 376(6592), eabk0639. https://doi.org/10.1126/science.abk0639

Patronek, G. J., Sacks, J. J., Delise, K. M., Cleary, D. V., & Marder, A. R. (2013). Co-occurrence of potentially preventable factors in 256 dog bite-related fatalities in the United States (2000–2009). Journal of the American Veterinary Medical Association, 243(12), 1726–1736. https://doi.org/10.2460/javma.243.12.1726

Sacks, J. J., Sinclair, L., Gilchrist, J., Golab, G. C., & Lockwood, R. (2000). Breeds of dogs involved in fatal human attacks in the United States between 1979 and 1998. Journal of the American Veterinary Medical Association, 217(6), 836–840. No DOI has been assigned to this work.

Serpell, J. A., & Duffy, D. L. (2014). Dog breeds and their behavior. In A. Horowitz (Ed.), Domestic Dog Cognition and Behavior (pp. 31–57). Springer, Berlin. No DOI has been assigned to this book chapter.

Frequently asked questions about “dangerous breeds”

What you should know about breed lists, prejudice, and behavior