How to End a Dog Training Session the Right Way
Start here
End every session on something your dog can already do. Ask for one easy known behaviour, mark it, pay it, release your dog, and put the treats away. Never stop in frustration straight after a failed repetition, and never cut a session short as a correction. The last thirty seconds are what your dog carries into the next session.
Why the last rep is the one your dog carries into tomorrow
A training session is not an average. Your dog does not walk away holding the mean of twelve repetitions. They walk away holding the last one, because that is what was happening when the food stopped, when your voice changed, and when you stood up and went back to your day. The ending is the part of the session with the clearest emotional label on it.
That matters for a practical reason. The next time you pick up the same skill, your dog arrives with whatever expectation the last session left behind. A dog who finished a recall session by nailing one short easy call in the hallway comes back to recall looking interested. A dog who finished the same session by ignoring three long calls in the park, then being clipped back on the lead and marched to the car, comes back having practised ignoring you.
Be honest about what kind of claim this is. Nobody has run a controlled trial isolating the final repetition of a session and measuring its effect a day later. What exists is a long-standing rule of training craft, worth following because it costs about twenty seconds.
Law 10: end each session on a high note
The rule comes from Karen Pryor's ten laws of shaping, set out in Don't Shoot the Dog! (Bantam, 1985, pp. 54-59). Law 10 is stated plainly: end each session on a high note. Pryor is writing about shaping across species, from dolphins to horses to people. Law 1 says raise criteria in small enough steps that the subject always has a chance of being reinforced. Law 9 says that if a behaviour falls apart, go back to kindergarten. Law 10 is the one about how you leave the room.
A high note does not mean a spectacular note. It means a repetition your dog completed successfully and got paid for. The easier the behaviour, the more certain the high note, which is why the right closing behaviour is almost never the thing you spent the session working on.
What that looks like in practice
Say you have spent four minutes teaching a puppy to hold a down while you take one step back. It is wobbly. On the sixth rep the puppy breaks the down and wanders off to chew the rug. Your instinct is to stop there, because the session is clearly over.
Instead, get the puppy back, ask for a sit they have done a hundred times, mark it, pay it, release them, and finish. Total cost: eight seconds. The session now ends on a behaviour performed correctly and reinforced, and the wobbly down gets picked up tomorrow from a neutral starting point.
Practical take: Close on a behaviour your dog gets right nine times out of ten, not on the skill you were actually training. The closing rep is a full stop, not another attempt.
This is also why session length and session endings are the same problem. Pryor puts the working ceiling at roughly a dozen reinforcements before appetite for the reward starts to fall off, with an upper bound around eighty a day. If you train until your dog is flat, there is no high note left to end on, because nothing is working any more. Stopping while your dog still wants more is what makes a good ending available. There is more on that in our guide to how long a training session should be.
The mistake almost everyone makes
Here is the pattern, and it is close to universal. You are having a good session. You get greedy and raise the difficulty one step too far. Your dog fails. You try again, a little tighter in the voice. They fail again. You feel the irritation arrive, decide you are being unfair to the dog, and stop.
Stopping there is the right instinct arriving at the wrong moment. You have just ended the session on the failure, and worse, you have ended it on the failure plus your frustration. The last thing your dog experienced was a repetition they could not solve while the person holding the food got tense.
Frustration is not a neutral ingredient. In Vieira de Castro et al. (PLOS ONE 15(12), 2020), ninety-two companion dogs were observed across aversive, mixed and reward-based training schools; the aversive group showed a mean of 55.9 lip-licks against 4.1 in the reward-based group, an effect size of d = 3.9. That is a study of training method, not of one tense minute at the end of a session, so do not over-read it. It is still a reminder that dogs read tension in the handler long before they read the training plan.
Practical take: When you notice the irritation, do not stop. Drop two steps back to something easy, get one clean success, pay it, and then stop.
Law 8: an abrupt stop is itself a punishment
Pryor's Law 8 says: don't interrupt a session gratuitously, because that in itself is a punishment. It is the law owners break without ever intending to punish anything. The dog gets silly, jumps up, grabs the treat pouch, barks at the neighbour, and the handler says right, we're done, and walks off with the food.
From the dog's side, that sequence reads cleanly. Good things were happening. I did something. Good things stopped. That is the textbook shape of negative punishment, and if you had designed it deliberately you would call it a time-out. Doing it by accident is worse than doing it on purpose, because you have no idea which behaviour you just suppressed. Often it is the offering of behaviour itself, which is the one thing you cannot afford to lose in a dog you are shaping.
Do not end a session as a way of telling your dog off. If a session has to stop because you are out of time, out of patience, or the dog is over-threshold, buy the ending first: one easy behaviour, mark, pay, release. Ending as a correction teaches your dog that working with you is a thing that can be taken away.
The broader evidence points the same way. In a controlled trial by China, Mills and Cooper (Frontiers in Veterinary Science 7:508, 2020), sixty-three dogs in three matched groups recalled on the first cue 82.5% of the time under reward-based trainers against 71.0% under trainers using electronic collars, with sit latency of 1.36 s against 1.67 s. That is obedience work under professional trainers, not a study of session endings, but it belongs to the same story. Our guide to positive reinforcement dog training goes through that evidence in full.
A three-line script for wrapping up any session
Use the same closing sequence every time, so it becomes automatic for you and predictable for your dog.
- One easy known behaviour. Sit, touch, hand target, a name response. Something with a very high success rate, asked once, in an easy position, with no added distance or duration.
- Mark and pay. Your marker word or click the instant the behaviour happens, then the food. This is a good place for a slightly better treat than the ones you were using, without making it a ceremony.
- Release word, then put the treats away. Say your release, all done or free, then physically put the pouch away. The order matters: the release ends the work, the pouch going away is just housekeeping after it.
Two honest caveats on the marker. Gilchrist et al. (PeerJ, 2021) ran two controlled studies in pet dogs on simple tasks and found a clicker gave no advantage over food alone or over a verbal marker, so for an easy closing rep the word in your mouth is fine. On harder work, markers earn their keep: Lazarowski et al. (Frontiers in Veterinary Science, 2025) randomly assigned twenty-eight detection dogs to marker or no marker on a difficult task with a long delay before reward, and the marker group reached the final stage in 36 trials against 68, alerting accurately 92% of the time against 69%.
If you want a bigger finish occasionally, Pryor's jackpot (p. 31) is an occasional out-of-proportion reward. Occasional is the operative word. A jackpot at the end of every single session is just your normal rate of pay, and it stops meaning anything.
When the session is already going badly
Sometimes there is no good rep to end on because the whole session has come apart. The dog is over-aroused, or the environment got harder than you planned, or they are simply tired. Law 9 covers this: if a behaviour deteriorates, go back to kindergarten. In practice that means dropping the criteria far enough that success is nearly guaranteed, and accepting that the last minute of this session is now about the ending rather than about progress.
Drop distance to nothing. Drop duration to nothing. Move to the easiest spot in the room, or the hallway, or indoors. Ask for a behaviour the dog offers freely. Take one success and finish.
If the dog cannot do a behaviour they normally offer without hesitation, in an easy place, when they are not obviously exhausted, that is worth paying attention to rather than training through. A sudden loss of a solid behaviour can be pain, ears, eyes or a general feeling of being unwell, and it is worth a vet check before you decide it is a training problem.
What a good ending buys you tomorrow
The point of all of this is not the twenty seconds at the end. It is the first thirty seconds of the next session. A dog who has ended fifteen sessions in a row on a success, paid and released, comes back to training keen, offering behaviour, watching you for the next thing. That keenness is the raw material everything else is built from, and it is quietly spent every time a session ends in a mess.
You do not have to be a good trainer to end well. You have to remember, which mostly means deciding on your closing behaviour before you start, so you are not choosing one while annoyed.
If you would rather not hold that structure in your head, Tavi's training module lays each session out for you, closing rep included, so the end of the session is already decided before you start it.
Every Tavi session ends on a rep your dog gets right
Tavi's training module orders skills by your dog's breed, age and how much training you have done before, so each session has a plan and a planned ending. Every rep is illustrated, and sessions run a few minutes rather than a long block. Free for your first pet.
Frequently asked questions
How should I end a dog training session?
End on a behaviour your dog already knows well. Ask for one easy repetition, such as a sit or a hand touch, mark it the moment it happens, pay it, give your release word, and then put the treats away. The whole ending takes about ten seconds. Do not end on the skill you were actually working on, because that is the one most likely to fail, and do not end straight after a failed repetition or while you are frustrated.
What does ending on a high note actually mean?
It means the last repetition of the session was performed correctly and reinforced. The phrase comes from Karen Pryor's Law 10 in Don't Shoot the Dog! (Bantam, 1985), the tenth of her ten laws of shaping. It does not mean the dog did something impressive. An easy sit that earns a treat is a better high note than a difficult behaviour attempted three times and half-achieved, because certainty is the whole point. Pick the closing behaviour for its reliability, not its difficulty.
Is it bad to stop training when my dog misbehaves?
Usually, yes. Pryor's Law 8 says do not interrupt a session gratuitously, because the interruption is itself a punishment. When you walk off with the food because the dog got silly, the dog experiences good things stopping, and you have no reliable way of knowing which behaviour you just suppressed. Often it is the dog's willingness to offer behaviour at all. If you have to stop, buy the ending first with one easy repetition, mark, pay and release.
Should I give a jackpot at the end of every session?
No. Karen Pryor describes a jackpot as an occasional, out-of-proportion reward, and the occasional part is what makes it work. If every session closes with a handful of chicken, that handful becomes the normal rate of pay and stops carrying any information. Use a slightly better treat for the closing repetition if you like, keep true jackpots rare and unpredictable, and save them for a genuine breakthrough rather than for the fact that the session is over.
My dog fell apart halfway through the session. What now?
Go back to kindergarten, which is Pryor's Law 9. Drop the criteria until success is nearly guaranteed: no distance, no duration, easiest available location, a behaviour your dog offers freely. Take one clean success, pay it, release, and finish there. Accept that this session is now about the ending rather than about progress. If a dog cannot perform a behaviour they normally offer easily, in an easy place, when they are not obviously tired, it is worth a vet check rather than more training.
Tavi's guides are informational and not a substitute for professional veterinary advice, diagnosis or treatment. When in doubt, call your veterinarian.
Sources
- Karen Pryor, Don't Shoot the Dog! (Bantam, 1985), pp. 31, 54-59 and 65
- China, Mills & Cooper, Frontiers in Veterinary Science 7:508 (2020)
- Vieira de Castro et al., PLOS ONE 15(12) (2020)
- Gilchrist et al., PeerJ (2021)
- Lazarowski et al., Frontiers in Veterinary Science (2025)