How to walk a big dog past another dog without losing it

Two big dogs meet on a narrow footpath. Both are on lead. Both handlers are trying to look casual. One dog stiffens, the other locks eyes. The lead goes tight. Someone lunges, someone yells, someone gets pulled sideways. Everyone leaves the walk shaken and the same thing will happen tomorrow, because nothing about tomorrow will be different.

Leash reactivity in a 40 kg (88 lb) dog is a physical incident, not a manners problem to solve later. It is also not aggression in the sense that most owners fear. Most reactive dogs are frustrated, over-aroused, or scared, not hunting. The distinction matters, because the fix for frustration is not the fix for fear, and neither one is punishment. Reactivity is a nervous-system problem. You cannot correct a nervous system out of a nervous system.

First, the word that changes everything: distance

Every dog has a distance at which the sight of another dog goes from information to alarm. Trainers call this the threshold. Under threshold, the dog can still hear you, take food, break eye contact, move. Over threshold, they cannot. The brain has left the building and there is no learning to be had until it comes back. Everything below rests on that one distinction: under threshold, work is possible; over threshold, only management is.

For most big dogs the threshold on a residential street is somewhere between five and twenty metres (roughly fifteen to sixty-five feet). It shifts by day, by wind, by whether the other dog is on lead or off, by whether the other dog is staring. This is why the whole game is buying distance in real time. Cross the street early, not once you're already committed. Duck behind a parked car so the sightline breaks. Turn a corner and give it thirty seconds. On a narrow trail with nowhere to go, step off into the grass, ask for a sit, and body-block the sightline yourself. None of this is failure. It is the work.

Under threshold, then engagement

Three signs the dog is still available: they can take a treat you offer, their ears are neutral or forward and mobile, and they can voluntarily look away from the other dog and back at you. If any one of those fails, you are too close. Add distance until all three come back. Do not try to force the third one. The dog voluntarily looking at you is the whole point, not something you can command your way into.

Once the dog is under threshold, the drill is a pattern game called 'look at me' or 'engage-disengage'. The dog sees the other dog, marker word ("yes"), treat delivered at your side. The dog is being paid not for good behaviour but for the specific act of noticing the trigger and choosing to check in. Six to ten reps in one session is plenty; more than that and the dog is doing the drill instead of learning it. The parallel structure is the same one that works at the front door for arousal. The big-dog jumping protocol runs on identical logic, just with a different trigger.

Currency matters early. Kibble is not enough for most dogs when a strange labrador appears across the street. Use something the dog would fight another dog for: cooked chicken, small cubes of cheddar, freeze-dried liver. This is not bribing. Bribing is producing the food before the behaviour to lure the dog into compliance; paying is delivering the food after the behaviour to reinforce it. The distinction is the difference between an unreliable trick and an actual reflex. As the reflex builds over weeks, the currency can drop back down to normal treats and eventually verbal praise. Not before.

The four-step protocol

Step 1 is walk at threshold distance only. For the first week, do not walk anywhere that will force you closer than the dog can handle. Change the route. Drive to a quieter park. Walk at 6am. This is the hardest ask in the whole protocol because it feels like giving in, but every over-threshold rehearsal sets the training back a week. If handling the dog at threshold is a struggle in itself, sizing the harness properly before the training starts is a real time-saver. A poorly-fitted harness on a barrel-chested dog rides up on the trachea and makes every moment worse.

Step 2 is engagement games at that safe distance. The look-at-me drill above, plus a hand-target (dog boops your open palm with their nose for a treat), plus a simple sit-and-watch. The dog learns that strange dogs in the environment predict fun for them. This is classical conditioning stacked on top of the operant reps; both are firing in parallel. Ten minutes a session, twice a day, is a solid dose. More than that is diminishing returns and a tired dog.

Step 3 is narrow the gap, one metre at a time. Once the dog is winning at twenty metres, try seventeen. Then fifteen. Then twelve. The pace is set by the dog: as long as the three under-threshold signs hold, you keep the new distance. The moment any one of them fails, you go back to the last distance that worked and stay there for another week. This is why the typical timeline is six months, not six weeks. Most owners rush this step and blow up the whole build.

Step 4 is the reset when it fails, because it will. The dog goes over threshold. Do not correct. Do not repeat the cue five times. Turn and walk the other way calmly, break the sightline, give the dog thirty to sixty seconds of just movement, and reset at a distance that works. That whole episode is data, not failure. The data says the distance you chose was wrong. Adjust and continue.

Gear that helps, and gear that quietly doesn't

A front-clip harness with a chest attachment is genuinely useful for reactive-dog walks. The physics are simple: a front-clip turns the dog sideways when they lunge, which interrupts the lunge before it commits. It does not train anything on its own; the training still has to happen. It just makes the management moments survivable while the training builds up. The Ruffwear vs Rabbitgoo comparison walks through which front-clip fits which shape of dog. A short, non-retractable lead in the four-to-six-foot range (1.2 to 1.8 m) belongs with the harness. A retractable at full extension puts a big dog fifteen feet away from you when the trigger appears, which is the exact wrong outcome.

What quietly does not help: prong collars and e-collars. Both suppress the outward behaviour without addressing what the nervous system is doing underneath. A dog that has learned to fear a correction still wants to lunge at that labrador; they have just learned to hide it, and the underlying arousal keeps building until something breaks. Modern behaviour research does not support aversive tools for reactivity in pet dogs, and the ways this fails tend to be worse than the way it started. Same story for the head halter used as a correction. A fine loose-lead tool in the right hands, not a reactivity fix.

How this usually falls apart

Three failure modes cover most of what goes wrong. The first is inconsistent distance: the handler holds threshold on Monday and Wednesday, and on Friday the neighbour's dog surprises them at three metres and it all unravels. The second is one household member running the drill correctly and the other letting the dog pull straight up to strange dogs on their walks. The third is training in a vacuum: an hour of drills in a car park with no strange dogs present, which teaches the dog nothing about the actual trigger.

There is a fourth failure mode worth naming on its own, because it looks like effort but does damage: flooding. This is the well-meaning owner who reasons that if the dog needs to get used to other dogs, more exposure must be better, so they take the reactive dog to the dog park and hope. The dog spends the hour over threshold, learns nothing except that other dogs mean panic, and the reactivity gets worse. Exposure without under-threshold work is not desensitisation; it is rehearsal of the exact problem you are trying to fix.

How long this actually takes

The floor is six weeks of daily practice for a young, mildly-reactive dog with a fully-committed household. That is the fast case. For an adult rescue with a longer history, the timeline is typically six months, sometimes twelve, and the endpoint is often not "my dog loves other dogs" but "my dog can walk past another dog without losing it," which is where this article got its title. A well-exercised dog is far more trainable than a bored one, so read our German Shepherd exercise piece for the arithmetic if you're not sure your dog is getting enough. Under-exercised big dogs walk into every training session already running hot.

When to bring in a certified behaviour professional (IAABC, CPDT-KSA, KPA-CTP, or an equivalent in your country): if the dog has bitten or has redirected onto a person, if you are afraid to walk them, if you have done the six weeks and seen no movement at all, or if the reactivity is bundled with resource guarding, separation issues, or a bite history. A pro will not do the work for you; the household still runs the daily reps. But they will spot the specific mechanical detail that has been quietly wrecking the drill for weeks. Money well spent.

There is no shortcut and no gadget. Only distance, currency, and the patience to run the same six-minute drill twice a day for as long as it takes. The dog you get at the end of it is not a different dog; it is the same dog with a nervous system that has learned another labrador on the far pavement is a normal part of the world. A very quiet kind of win, and the only kind that lasts.