Department of Justice Canada reported in June 2026 that 9,532 of the 21,218 adult criminal court cases potentially at risk of exceeding the Jordan timelines in 2023/2024 were stayed or withdrawn, and that the reason cannot be determined because the information is not currently reported to Statistics Canada by the courts

What is happening?

In July 2016 the Supreme Court of Canada set numerical ceilings for delay in criminal proceedings in R. v. Jordan: 18 months for a case tried in provincial court and 30 months for a case tried in superior court, or in provincial court after a preliminary inquiry. Past the ceiling the Crown must establish exceptional circumstances. The Justice Canada Charterpedia entry on section 11(b) of the Canadian Charter of Rights and Freedoms describes a stay of proceedings as the minimum remedy for a breach, with that page current to 2025-12-31.

Statistics Canada counts how often cases come close to those ceilings. Using the Integrated Criminal Court Survey it publishes an indicator of cases potentially at risk of exceeding the Jordan timelines, measured from the first court appearance to the final decision.

Department of Justice Canada published a report in June 2026 covering 2016/2017 through 2023/2024. In 2023/2024 there were 228,425 completed adult criminal court cases, of which 21,218, or 9 percent, were potentially at risk of exceeding the Jordan timelines. Of those at-risk cases, 9,532, or 45 percent, were stayed or withdrawn. In 2016/2017 the comparable figures were 355,951 completed cases, 20,619 at risk, or 6 percent, and 5,724 at-risk cases stayed or withdrawn, or 28 percent.

The report then states that it is not possible to determine the reason for which those cases were stayed or withdrawn, because that information is not currently reported to Statistics Canada by the courts. Two Statistics Canada articles, one published in 2020 and one on 2026-03-24, record the same limitation from the survey side: the survey cannot attribute elapsed time to the Crown, to the defence or to institutional causes, which is why the published figure is described as a maximum rather than a count.

So the ending is counted and the reason for it is not. A stay entered because a court found the delay unreasonable and a withdrawal entered for any other reason arrive in the national statistics as the same value.

Whose problem is this?

The body that publishes the indicator and the body that names its limitation are the same body. What is missing is not an analysis and not a finding but a field on a reporting form.

Where does this problem end?

The boundary here is the reporting field and not the remedy, because whether a stay is the right answer to delay is a question this medium does not settle. A reader who wants the second question answered will not find it here, and that is deliberate.

What is the state now, and what should it be?

Now

The two rankings in the last two rows do not share a denominator. The first is a share of all completed cases in a jurisdiction and the second is a share of the at-risk cases in a jurisdiction, and a province can rank high on one and not on the other.

The count of endings exists in the national statistics and the reason for those endings does not.

Needs a new measurementthe target state: no source opened here names a requirement that the reason for a stay or withdrawal be reported, a date by which such reporting would begin, or a target for how often delay should end a case. The June 2026 report recommends continued monitoring of the trend, which is a recommendation about watching the existing indicator rather than a target for completing it. The 2026 statute states a purpose of reducing the number of cases abandoned because of the timelines, and no source opened here attaches a number or a date to that purpose.

How big is it?

21,218 cases in 2023/2024 were potentially at risk of exceeding the Jordan timelines, and 9,532 of them were stayed or withdrawn. Both figures come from the June 2026 Department of Justice Canada report, which pairs them and describes the second as 45 percent of the first.

The comparison across years has to be read with care. In 2016/2017 there were 20,619 at-risk cases and 5,724 of them were stayed or withdrawn. The number of at-risk cases barely moved between the two years, from 20,619 to 21,218. The share of completed cases that were at risk moved from 6 percent to 9 percent, and over the same span the count of completed cases fell from 355,951 to 228,425. The two years do not share a denominator, so the count and the share do not tell the same story and neither one should be reported as the other.

The unit reported in every source opened here is the case and not the person. No source gives a number of accused persons, victims or witnesses attached to these cases, which is why this document carries no population estimate.

And the 9,532 is a ceiling, not a measure of delay. A case can be stayed or withdrawn for reasons unrelated to elapsed time, and the report states that the statistics cannot tell those apart from a stay entered because the delay was found unreasonable. The one jurisdiction that publishes its own delay applications posted 52 filed between 2025-04-01 and 2026-03-31. That figure and the national at-risk count are not the same quantity, because one counts applications made in a single province and the other counts cases flagged across the whole country, and no source opened here bridges them.

Under what conditions does it arise?

1. The survey already receives the record and does not carry the field. The Integrated Criminal Court Survey takes completed adult criminal court cases and computes elapsed time from the first court appearance to the final decision, which is how the at-risk indicator exists at all. The outcome category records that a case was stayed or withdrawn. What is absent is one attribute of that outcome, so the correction surface is an existing reporting item rather than a new collection system.

2. The measured quantity is defined as an upper bound and reads as a count. Statistics Canada records that the survey cannot attribute elapsed time to the Crown, the defence or institutional causes, and that the case clock starts at first appearance rather than at charge, so the figure may overstate. A number published as a maximum circulates as an estimate once it leaves the page that qualified it.

3. Publication is not uniform across the country. The federal ombudsperson office records that access requests to the thirteen provinces and territories produced no pan-Canadian figure and inconsistent reporting frameworks, and that one jurisdiction publishes counts of delay applications on its own initiative. A national picture assembled from thirteen different frameworks is not the same measurement thirteen times.

4. The remedy rule changed while the frequency of the remedy remained uncounted. The Protecting Victims Act received royal assent on 2026-06-18 and directs courts to order a stay only where satisfied that no other remedy is appropriate and just. The rule that fires was rewritten without a national count of how often it fired before.

What has been tried?

Every attempt in this list either measures elapsed time or changes what happens when elapsed time is excessive. None of them adds the reason for the ending to what the courts report.

What was found?

Why is it still unsolved?

Measurement absent — the record that would answer the question is already in transit, and the answer is not one of the fields it carries.

A number that is measured at its upper bound tells you how large a problem could be and never how large it is. That is the whole distinction between the indicator that exists and the indicator that does not. The survey knows a case took longer than the ceiling and it knows the case ended in a stay or a withdrawal. It does not know whether those two facts are related. Every reading of the 9,532 therefore has to carry a silent maximum sign, and the maximum sign is the first thing that falls off when a figure is quoted.

The second part is that the gap does not look like a gap. A missing programme is visible because nothing happens. A missing field is invisible because the report still publishes, the indicator still has a value, the trend still has a direction, and the annex table still carries a value for every year. The June 2026 report is honest about the limitation in the same paragraph that carries the number, and the limitation does not travel with the number afterward.

The third part is that the rule which depends on the count was changed while the count stayed unavailable. The 2026 statute narrows when a stay may be ordered and tells courts to consider other remedies first. Whether that changes anything is a question about how often stays are entered for delay, and that is precisely the quantity no national source reports. A rule can be amended and then evaluated against the same indicator it was already impossible to evaluate against, and nothing in the record marks that as a failure.

What observation would mean it is solved?

Candidates — (a) the statistics carry the reason, so that stays entered for unreasonable delay are separable from stays and withdrawals entered for other reasons (b) a pan-Canadian count of applications and stays under section 11(b) is published on a regular schedule (c) the share of cases potentially at risk of exceeding the timelines falls.

(a) is the one that matches the gap, and it is still not sufficient alone. A reason field can be added and filled inconsistently across thirteen reporting frameworks, which would produce a national figure that is the sum of thirteen different definitions. Coding practice would have to be comparable before the resulting number means anything, and no source opened here describes what such a coding standard would contain.

(b) counts a different thing. Applications are filed by the defence and stays are ordered by a judge, and neither quantity equals the number of cases that ended because of elapsed time. The one jurisdiction that publishes applications posted 52 for a single year, and its own breakdown shows that some were granted, some dismissed, some abandoned and some resolved for unrelated reasons, so even a complete national application count would describe the filings rather than the endings. A national application count would be a genuine improvement and would still leave the outcome field empty.

(c) can move without anything improving. The at-risk share is a ratio, and between 2016/2017 and 2023/2024 its denominator fell from 355,951 to 228,425 while its numerator stayed nearly flat at 20,619 and then 21,218. A share computed on a shrinking caseload can move because fewer cases are being completed rather than because cases are moving faster. The three have to be read together, and (a) is the only one of them that makes the other two interpretable.

What is it connected to?

Fills with researchthe number of people attached to each case, the relationship between judicial and prosecutorial capacity and elapsed time, court backlog measurement in other countries, and the effect of the 2026 remedy amendments once they have been in force long enough to observe. Relation type and evidence grade were not established in this round.

What these sources do not say

• Why an at-risk case was stayed or withdrawn. The June 2026 federal report states that this cannot be determined because the information is not currently reported to Statistics Canada by the courts, and the two Statistics Canada articles record the same limitation. No source opened here supplies the missing attribution by another route.
• How many stays were entered under section 11(b) nationally. No page opened here carries that figure. That is a statement about what these sources contain and not a demonstration that no such count exists anywhere.
• Whether the 2026 delay amendments carry any data collection or reporting obligation. The summary carried with the bill and the federal department page describing the statute do not mention one, and the enacting text of the new delay part and its coming-into-force section could not be read in full. The parliamentary legislative summary would be the place to check, and that document could not be read either.
• Whether provincial ministries hold counts of delay stays and do not publish them, or do not compile them. The federal ombudsperson page records that access requests produced inconsistent reporting frameworks and no pan-Canadian dataset, and it does not distinguish between those two situations. Neither does any other source opened here.
• Why the published breakdown of one provincial count does not reconcile. The posted categories for the 52 delay applications sum to 51 as read, and the page carries no note explaining the difference.
• What recommendations the federal ombudsperson office made. The recommendations section of the web version does not contain a data collection recommendation, and the report version of the same document could not be read, so whether one exists there is unresolved.
• How the at-risk indicator relates to the survey definition of a final decision category. The definitions section of the June 2026 report was not read, so how the stayed and withdrawn category interacts with the at-risk flag is described here only as the report presents it.
• Any on-the-record explanation from the courts or from provincial justice ministries. No statement from either was found in any source opened here. Department of Justice Canada, Statistics Canada and the one province that publishes proactively all document the limitation themselves, and the bodies that would supply the missing field are the ones whose position could not be located.

See the evidence

No court record and no enacting statutory text was read in full. The figures of 21,218 at-risk cases and 9,532 of them stayed or withdrawn in 2023/2024 come from a single federal report opened directly, which publishes the two together and states the limitation in the same passage. That limitation is corroborated from the other side by two Statistics Canada articles written by the agency that runs the survey, so the central claim of this document rests on two independent federal sources that agree. The 18 and 30 month ceilings appear identically on three federal pages, which is why they are treated as settled even though the 2016 judgment itself returned HTTP 403 to every attempt. Where a source is secondary it is marked so in the findings table, and no claim carried in the opening of this document rests on a secondary source. Where a document could not be read the row is kept with the URL column blank and the reason stated, rather than removed or filled with a substitute address. This is a Path A output, a research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below