Proving Coercive Control After Ahluwalia

The pattern was always the thing everyone in the room could feel and nobody had a place to put.

Courts wanted incidents. What happened, on what date, and can you prove it. So the work I do — reading a record, and saying whether there is a pattern in it, what the pattern is, and what it did to the person living inside it — produced an answer the law had limited use for.

That has changed, and it has changed quickly. Below is what changed, and — more usefully for anyone with a file in front of them — what it now asks a person to be able to show.

Three changes, and only two of them are live

Almost everyone working in this area knows something changed. Far fewer can say what. They do quite different jobs, they arrived at different times, and one of them is not in force yet. This is the single most common muddle in the subject, and it is the one most likely to be exposed later, in a report or in a witness box.

Two things about the Divorce Act definition do a great deal of quiet work. The behaviour does not have to be a crime to count — nothing needs to have been charged and nobody needs to have been convicted. And a child does not have to be hit, or even in the room. Exposure counts, and indirect exposure is the part most often left out: the child who could hear it through a wall, or who learned to read a parent's mood from the sound of the front door.

The criminal offence is the one people most often get wrong. Coercive control cannot be charged today. Parliament passed it in June 2026 and then gave the system up to two years before it comes into force, because the Department of Justice was explicit that the time was needed to train people. The gap is deliberate.

So at the moment the family and civil doors are open and the criminal one is not. That is counter-intuitive — the instinct is to treat criminal law as the real law — but on this subject, right now, it is the other way round.

What the Supreme Court actually said

On 15 May 2026, six judges to three, the Court recognised a tort of intimate partner violence. It applies across Canada, and in a family case it can be run inside the proceeding rather than as a separate lawsuit.

Three things have to be shown, on the balance of probabilities — more likely than not, the ordinary civil standard. Nothing has to be proved beyond a reasonable doubt.

1. The conduct arose in an intimate partnership, or in its aftermath.
2. The person acted intentionally, not accidentally.
3. Assessed objectively and in context, the conduct amounts to coercive control — not one incident standing alone, but the pattern taken as a whole.

Four parts of the decision matter enormously to the people I work with, and I want to set them out plainly because they are frequently misunderstood.

No separate injury has to be proved. The Court held that the harm flows from proof of the conduct itself. There is no requirement to produce a diagnosis. I cannot overstate how much this matters. For years the practical effect of the old approach was that a person had to be damaged enough, and be able to document being damaged enough, before anyone would look at what had been done to them. That incentive is gone.

Physical violence is not required. The pattern can be made entirely of things that would never trouble a police officer — who she sees, what he says about her to her mother, whose name the account is in.

Conduct after separation counts, including through the courts. Control does not stop when the relationship ends; it changes arena. The Court named misuse of the court process as one of the ways it continues, which is a thing many people have lived and few have been believed about.

Damages have to mean something. The Court said the amount must be a meaningful response to the seriousness of what happened, not a nominal gesture.

Why the pattern is hard to show

Here is the practical difficulty, and it is the whole of this work in one sentence: the law asks about a pattern, and almost every record anyone will hand you is organised around incidents.

Police occurrences are incident-based. Disclosure is incident-based. Clinical notes are appointment-based. Messages arrive as an undifferentiated wall. And each item, held up on its own, is almost always unremarkable. A text asking where she is. An account moved into one name. A cancelled appointment. A late pickup.

Taken one at a time, in the order they are usually presented, a pattern can be dismantled into a series of trivialities in about twenty minutes. I have watched it happen.

The work is not finding the one decisive document. There usually isn't one. The work is assembling a record in which the shape is visible without anyone having to take one person's word for it.

What I look for

• Sequence before content. I put everything on a timeline before I read any of it closely. Conduct that clusters around a separation date, a court date, a school transition or a holiday means something that the same conduct scattered at random does not.
• Records made by people with no stake in it. Banks, employers, schools, pharmacies, housing and immigration files. These are the hardest to wave away as one side's version of events.
• Volume and timing in communications, not wording. What any single message says is usually the least informative thing about it. How many there were, at what hour, in answer to what, and what the other person had to rearrange because of them — that is where a pattern lives.
• The "so what" of every item. Evidence that isn't connected to an effect is not yet evidence of control. What did this person stop doing, stop having, or start managing around? Without that link, a list of grievances is just a list.
• What is about to be lost. Phones get wiped, accounts get closed, platforms expire data. A great deal of the best material in these files disappears before anyone thinks to ask for it.

The two ways to get this wrong

Both of them are real, and I have to be able to fall on either side.

The first is not seeing it — reading a pattern as a run of unconnected small events, or as two people who simply stopped liking each other. That is the error the law was changed to address, and it is the one that has done the most damage over the longest time.

The second is over-claiming, and it is not the lesser problem. Not every bitter separation involves coercive control. Situational conflict, high-conflict separation and coercive control are genuinely different things, and they get conflated constantly. Two people who are both aggressive and neither of whom is subordinated are not in a coercive relationship, however awful it is to be near.

I say this not as a caution to other people but as a description of my own obligation. If I stretch the concept to fit a file where it doesn't belong, I have not helped that person — I have handed the other side a reason to distrust everything else in the report, and I have made it harder for the next person, whose pattern is real, to be believed.

What I can say, and what I can't

I am a clinician, not a lawyer, and there is a line I do not go near: I cannot tell a court whether the tort is made out. That is the judge's decision, and an expert who edges toward it is doing damage, usually to the party who retained them.

What I can do is explain what a pattern is and how it works, apply a stated method to the record, and set out what that record does and does not support — including, when it is the honest answer, that it does not support the claim.

In Ontario the rules are clear about this. An expert's duty runs to the court, and it outranks any duty to whoever is paying. The Supreme Court has said an expert has to be willing and able to be fair, objective and non-partisan, and if a report reads as advocacy it will be given little weight — and the other side will notice long before the judge does.

Which is also why I take instructions from either party. Someone facing an allegation that does not hold up needs a careful pattern analysis every bit as much as someone whose pattern is real and has never been named.

Said plainly

This is young law. Three of the nine judges would not have recognised the tort at all. The majority is the law, but where the edges fall will be worked out by lower courts over years, not months — how the objective standard gets applied, what a meaningful award looks like, how this sits alongside the existing family law analysis.

Anyone who tells you the outcome of one of these cases is obvious is overselling. That includes anything you read here.

If this is live for you

• Expert reports and case consultation for family, civil and criminal proceedings, in Ontario and across Canada — for counsel on either side, or for someone representing themselves.
• Free resources — six questions for lawyers to ask before commissioning an expert, a briefing for clinicians whose notes may go to court, and a plain-language guide to this decision.
• Two online courses on recognising coercive control and on proving it, open to anyone; live training for organisations on request.

Elizabeth Joy Dubois, MSW, RSW. General information about a changing area of law and how it meets clinical work — not advice on your own matter.

What I've relied on here: Ahluwalia v. Ahluwalia, 2026 SCC 16; the Divorce Act and Ontario's Children's Law Reform Act; Criminal Code s. 264.01 as enacted by the Protecting Victims Act, which received Royal Assent on 18 June 2026; the Ontario Family Law Rules on expert evidence; White Burgess (2015 SCC 23) on an expert's duty; Leitch v. Novac on litigation abuse; and Kelly and Johnson's work on differentiating types of intimate partner violence. Current to August 2026.