Service Dog Task Training: Tasks and Timelines

Build each task as a chain the dog can perform on cue and then on a cue the dog reads from you rather than hears - the alert, the interruption, the retrieve, the brace - because under the ADA it is the trained task that makes a dog a service animal, and the task is the only part of the work that can actually be trained.

What this page covers, and what the other one covers

Our guide to training a service dog owns the law: the ADA definition, the two questions staff may ask, why documents sold online convey no rights, and the fact that training your own dog is legal. Read that first if the question is about access.

This page is about the work itself. What a task actually is, how psychiatric alerts, deep pressure therapy and mobility support are built, what the published research says about how reliable any of it is, and how long each takes.

They are different questions and they fail in different ways. Getting the law right and the tasks wrong produces a dog with rights it cannot use.

What counts as a task

A specific trained action, performed in response to something. The Department of Justice's line is that a service animal is a dog individually trained to do work or perform tasks for a person with a disability, and that the tasks must be directly related to that disability.

The illustration DOJ itself gives is the clearest test available. A dog trained to sense an oncoming anxiety attack and take a specific action to help avoid it or lessen its impact qualifies; a dog whose mere presence provides comfort does not.

So the question to ask of any candidate task is: what does the dog do, on what signal, and how would a stranger know it had happened? "Calms me down" fails that test. "Puts its chin on my thigh when I start bouncing my leg, until I take my hand off it" passes.

Psychiatric tasks, and what veterans actually rate as useful

Rodriguez and colleagues surveyed 217 post-9/11 US veterans with PTSD — 134 who had a psychiatric service dog and 83 on the waitlist — and asked how important each trained and untrained behaviour was, how often it was used, and which symptoms it helped.

Calming and interrupting anxiety came out as the most important trained tasks, the most frequently used in a typical day, and the ones that helped the most symptoms. Trained tasks were most useful against hypervigilance and intrusion, and least useful against amnesia and risk-taking.

Then the finding that ought to change how these dogs are marketed: veterans rated the dogs' untrained behaviours as, on average, more important for their PTSD than the trained tasks. And veterans still on the waitlist expected higher task use and importance than veterans who actually had a dog reported.

Neither of those makes task training pointless — the trained tasks are what the ADA requires and what the dog is legally for. They do mean that a programme promising the tasks will do the work is over-promising, and that the expectation gap is measurable.

Deep pressure therapy, honestly

Deep pressure therapy is the trained behaviour of applying body weight across a handler's lap, chest or legs on cue, and holding it. It is one of the most commonly trained psychiatric tasks and one of the most commonly asked about.

It is also one of the least evidenced. We could not find a controlled study of deep pressure therapy as a trained canine task, and anything you read that gives a success rate for it is inventing one. What exists is Rodriguez's ranking above, where calming behaviours score highly on perceived importance and daily use.

Two practical constraints are worth more than a claim. The dog must be a size and structure that can apply useful pressure without hurting itself, and the behaviour has to have a release cue trained as hard as the start, because a dog that will not get off is a dog that cannot be worked in public.

Alert tasks, and what "reliable" means

This is where the only good numbers in the field live. Rooney and colleagues assessed 27 trained glycaemia alert dogs against nearly 4,000 out-of-range blood glucose episodes using routine owner records.

Median sensitivity to out-of-range episodes was 70%, with a wide spread between dogs — 50% at the 25th percentile and 95% at the 75th. Sensitivity to hypoglycaemic episodes was better than to hyperglycaemic ones, at a median of 83% against 67%. Median positive predictive value was 81%, meaning roughly four in five alerts came when glucose really was out of range, and four dogs reached 100%.

The authors' conclusion is that performance is better than earlier small studies suggested and that it varies enormously by individual dog, partnership and household. Read those numbers as the honest ceiling for any alert task: useful, real, and not a substitute for a monitor or a medication schedule.

Mobility tasks and the dog's own body

Mobility work splits into two kinds and only one of them is about strength. Retrieval, opening doors, pulling laundry, pressing buttons and bringing a phone are all light tasks that most sound dogs of adequate size can do.

Counterbalance and bracing are not. A dog taking a person's weight is loading its own joints repeatedly, and doing it before skeletal maturity or on a dog with hip or elbow disease is a welfare problem with a bill at the end of it. Get orthopaedic clearance first, and match the task to the dog's size honestly rather than to what you would like it to be.

There is no federal standard specifying a dog-to-handler weight ratio for this work, and the ratios circulating online are convention rather than evidence. That is a gap we would rather name than paper over.

How a task is actually built

Six steps, identical whether the task is a nudge, a retrieve or a brace. AVSAB names the toolkit: capturing, shaping and luring, with a marker.

Define the finished behaviour in one sentence

Written down, including the release. If you cannot describe it in a sentence a stranger could score, you cannot train it consistently and neither can anyone helping you.

Split it into the smallest piece the dog already does

Chin near the leg before chin on the leg. Nose to a hand before nose to a hand for five seconds. The split is the skill; the rest is repetition.

Mark the instant, pay fast

A clicker or one consistent word, at the exact moment the criterion is met, followed by food quickly. The mechanics are the same ones in dog trick training, which is the cheapest place to practise them on a behaviour nobody depends on.

Add duration before you add the cue

Chain length first, name second. A cue attached to a half-built behaviour produces a dog that starts the task and abandons it.

Move the trigger from your cue to the real one

This is the step that makes a task a service task. For an alert, the dog is rewarded for responding to the physiological or behavioural sign rather than to your word — which means capturing what the dog already notices, not teaching it something new.

Proof it in the places it has to work

Empty room, busy room, shop, clinic, aircraft cabin. A task that only exists at home is not yet a task, and public-access behaviour is the other half of the job that the ADA guide covers.

How long does it take to train a service dog?

There is no federally required number of hours, because there is no federal certification — DOJ states plainly that entities may not require proof that an animal has been certified, trained or licensed. Anyone quoting you a legal minimum is quoting a private standard, not the law.

In practice, a single simple task takes weeks. A dog raised and trained from puppyhood to work reliably in public takes 18 months to two years, and a proportion of candidate dogs never qualify — which is the timeline our ADA guide uses and the one programmes plan around.

The tasks are rarely the long pole. Public-access behaviour is: being boring in a hundred places, for hours, next to food, children and other dogs.

Do psychiatric service dogs actually work?

The best evidence is promising and short of proof. O'Haire and Rodriguez ran a non-randomised trial with 141 post-9/11 military members and veterans, comparing usual care alone with usual care plus a trained service dog.

They found clinically significant reductions in PTSD symptoms after receiving a dog but not during usual care alone, plus lower depression, higher quality of life and higher social functioning against the waitlist group, with medium to large effect sizes. Average symptom reduction did not take people below the diagnostic cut-off, and the authors describe the work as proof of concept.

That is a real result and it is not a cure. Treat a service dog as a complementary treatment your clinician knows about, not a replacement for one.

What about comfort dogs and therapy dogs?

Different job, different law, and it has its own page. A therapy dog visits by invitation and has no public-access rights; a comfort dog usually turns out to mean either that or an emotional support animal, which is a housing category.

The distinctions and the qualifying route are in therapy dog training. If what you need is a dog that performs trained tasks for your own disability, you are on the right page — and the handling foundation that keeps it examinable is cooperative care training.

Related guides

• training dog as service dog — Training a dog as a service dog means task work for a disability.
• pet therapy dog training — A therapy dog visits by invitation and has no public access rights.
• dog trick training — Tricks are where you build marker timing, rate of reinforcement and shaping.

Every guide we have written is on the dog training guide index.

Sources

1. Frequently Asked Questions about Service Animals and the ADA — US Department of Justice, Civil Rights Division (ADA.gov), 2026.
2. ADA Requirements: Service Animals — US Department of Justice, Civil Rights Division (ADA.gov), 2026.
3. Defining the PTSD service dog intervention: perceived importance, usage, and symptom specificity of psychiatric service dogs for military veterans — Frontiers in Psychology 11:1638 (Rodriguez, LaFollette, Hediger, Ogata & O'Haire), 2020.
4. Preliminary efficacy of service dogs as a complementary treatment for posttraumatic stress disorder in military members and veterans — Journal of Consulting and Clinical Psychology 86:179-188 (O'Haire & Rodriguez), 2018.
5. How effective are trained dogs at alerting their owners to changes in blood glycaemic levels? Variations in performance of glycaemia alert dogs — PLOS ONE 14(1):e0210092 (Rooney, Guest, Swanson & Morant), 2019.
6. Position Statement on Humane Dog Training — American Veterinary Society of Animal Behavior (AVSAB), 2021.