Why Does My Dog React to Some Dogs But Not Others?

By Jessica Ennis, DVM, CVA

✓ Medically reviewed by Jessica Ennis, DVM, CVA — August 11, 2026 For general education — not a substitute for veterinary care.

Behavior · August 10, 2026 · 6 min read

Most dogs aren't reactive to "dogs" as a category — they're reacting to a narrow set of triggers, and finding that set is most of the fix.

People assume a reactive dog is reactive to dogs, full stop. That’s rarely how it actually works.

I’ve had more owners than I can count sit in my exam room convinced their dog “hates dogs,” only for it to come out over ten minutes of questions that the dog is fine at the park, fine with the neighbor’s beagle, and only loses it at intact male dogs, or big black dogs, or dogs that stare. That’s not a temperament problem. That’s a pattern, and patterns can be worked with.

One dog I worked with, a four-year-old spayed border collie named Pepper, thirty-eight pounds, came in because her owner was ready to give up on off-leash hikes. Pepper was easy around dogs she met loose and moving — greeted, sniffed, moved on. But on leash, at a dead stop, facing a dog head-on, she’d go rigid, stare, and start a low bark that built fast. The owner had tried three different “calming” chews with no change. What actually mattered wasn’t the chew — it was that every problem encounter was leashed, frontal, and stationary. Off that specific setup, Pepper was a different dog. Once we changed the geometry of greetings — angled approach, kept moving, more distance — the barking dropped out almost entirely within a few weeks.

What “some dogs, not others” usually means

Selective reactivity is almost always about a specific set of variables, not the whole species. The most common ones I see, in rough order of frequency: size, coat color and pattern, ear/tail carriage, sex and reproductive status, breed-typical body shape, and behavior style — a stiff, staring dog reads very differently to a nervous dog than a loose, bouncy one does. Dogs are extremely good at pattern-matching, and a dog that had one bad encounter with, say, a large black dog with erect ears can generalize that fear or arousal to anything that resembles it, even years later. This isn’t reasoning in the way we’d think about it — it’s associative learning, and it’s fast, especially after a single strong experience.

It’s also worth separating what looks the same from a distance but isn’t. A dog straining and barking at every dog on the other side of a fence is often showing barrier frustration, not aggression — arousal about being blocked from getting to something, which reads differently once the barrier’s gone. A dog that goes still, leans back, and shows a hard stare before another dog even approaches may be communicating something closer to what’s described as whale eye — visible eye-white paired with a stiff body, a warning sign worth reading rather than a random quirk.

Why triggers cluster the way they do

Body language mismatch is the biggest driver I see. A calm, loose dog approaching another calm, loose dog is usually fine. Add a stiff-postured, direct-staring dog into that mix and tension goes up fast, because stillness and direct eye contact read as a threat signal in dog body language, not a neutral one. Dogs who were under-socialized to a particular look — say, never around brachycephalic breeds, or never around dogs with cropped ears — sometimes react less out of learned fear and more because the shape genuinely doesn’t parse as “dog” to them at first glance. That’s a real phenomenon, and it tends to resolve faster with calm, controlled exposure than the fear-based version does.

Prior experience matters enormously too. A single bad encounter, especially during a fear period in puppyhood, can create a lasting association with a specific look or size class. This is single-event learning, and it doesn’t need repetition to stick.

Ruling out what isn’t really about the other dog

Before assuming this is purely a social issue, it’s worth ruling out pain. A dog who’s sore — arthritis, a healing injury, dental pain — has a lower bar for defensive reactivity, and that irritability can look exactly like “doesn’t like certain dogs” when it’s really “doesn’t want to be jostled right now.” I ask owners about recent changes in jumping, stairs, or willingness to be touched near the hips or spine before I treat a new reactivity pattern as purely behavioral. Age matters here too — the same logic applies to sound sensitivity, where an older dog can start reacting to noises he used to ignore simply because hearing, vision, or joint comfort changed, not because his temperament did.

It’s also worth distinguishing this from what’s happening at the door or on walks generally. A dog who’s fine with dogs but escalates around strangers, or who only struggles when guests arrive, is often dealing with something closer to territorial or frustration-driven arousal than dog-specific reactivity, and the intervention looks different.

The owners I see most often at this point have already tried generic advice — more exercise, a different leash, a calming chew — and are frustrated it didn’t move the needle. I remember a twelve-year-old golden retriever named Otis, sixty-eight pounds, whose owner was sure he’d “gotten aggressive” toward small dogs over about two months. What had actually changed was Otis’s hips — he’d started declining stairs and hesitating before jumping onto the couch that same window, and the small-dog encounters were the ones most likely to end in a fast, unpredictable approach he no longer had the comfort or mobility to sidestep calmly. Addressing the joint discomfort and giving him more room and time to disengage did more than any behavioral protocol would have on its own.

Where calming support genuinely fits, and where it doesn’t

Calming ingredients are not a fix for selective reactivity — they’re, at best, a way to lower baseline arousal so training has room to work. The evidence behind the common ones is real but uneven. A review on Biomedicines notes that for chamomile’s calming effects, “for the remaining species, evidence is derived mainly from preclinical studies or traditional use” — meaning the human and lab-animal data don’t automatically transfer to dogs. Ashwagandha has somewhat stronger backing: a systematic review in Cureus “analyzed multiple RCTs and concluded that ashwagandha has a moderate positive impact on stress reduction and cognitive performance,” though again, that’s evidence built mostly on human trials. L-theanine’s picture is more mixed — a review in Molecular Psychiatry found benefit “except for one study on psychotic anxiety (SMD = 0.54; 400 mg/day for 8 weeks),” a human dosing context that doesn’t map directly onto dogs. None of this is a substitute for identifying and managing the actual trigger pattern; think of it as, at most, one small lever alongside distance management and training, not the plan itself.

What actually changes the pattern

The most useful thing an owner can do is become a careful observer before anything else: log what the dog reacted to, at what distance, and what the other dog’s body language looked like. That log usually reveals the pattern within a couple of weeks — size, color, posture, sex, or some combination. From there, management (distance, angle of approach, avoiding head-on stationary meetings) does more in the short term than any single product, and controlled, calm exposure to the specific trigger category, done gradually and below the threshold where the dog reacts, is what actually shifts the association over time.

If the pattern is worsening, involves biting or near-misses, or shows up alongside other new behavior changes, that’s the point to bring in a trainer or veterinary behaviorist rather than trying to work through it alone.

The short version

Dogs that react to “some dogs but not others” are almost always responding to a specific, identifiable trigger — not dogs as a category. Track the pattern, rule out pain or an underlying medical shift, manage distance and approach angle during the process, and treat calming ingredients as a minor supporting player at most. The dog isn’t broken; the trigger just hasn’t been named yet.