Binding 12-hour and 48-hour deadlines for requests for services to First Nations children have stood since 2017 — as of 2026-08-15 no published rate of adherence to them was found in any source opened here

What is happening?

Jordan's Principle is a rule about who pays first. When a First Nations child needs a health, social or educational service and two governments or two departments disagree about which of them carries the cost, the government of first contact pays and the dispute over responsibility is settled afterwards. It is named after Child A, a five year old child from Norway House Cree Nation in Manitoba who died in hospital in 2005 while the federal and provincial governments disagreed over which of them would pay for his care at home. The House of Commons adopted a motion in support of the principle, M-296, unanimously on 2007-12-12.

The rule acquired legal force through a human rights case rather than through a statute. In January 2016 the Canadian Human Rights Tribunal held in 2016 CHRT 2 that the federal First Nations Child and Family Services program, and the narrow definition Canada was then applying to Jordan's Principle, amounted to discrimination under the Canadian Human Rights Act. In May and November 2017, in 2017 CHRT 14 and 2017 CHRT 35, the tribunal turned response times into binding orders: 12 hours for an individual urgent request, 48 hours for an individual non urgent request, 48 hours for a group urgent request and one week for a group non urgent request.

Nine years after that first ruling the pile of unanswered requests is measured in six digits and is not falling. About 140,000 requests were recorded as backlogged at 2024-12-04, of which about 25,000 had been marked urgent by the people who filed them. About 135,000 were reported at 2025-02-20, alongside a national appeal backlog of 727 and about 8,000 requests marked urgent. At 2025-06-04 the figure was again about 140,000, and about 100,000 new requests had arrived in the six months since December 2024. Around January 2026 a federal official confirmed in cross-examination before the tribunal that the national backlog stood at about 130,000.

The tribunal has already ruled on this, and Canada asked a court to undo the ruling. In 2025 CHRT 6, a summary decision issued 2024-11-21 with full reasons following on 2025-01-29, the tribunal granted a non compliance motion and ordered Canada to clear the backlog. On 2024-12-20 the federal government applied to the Federal Court for judicial review to set that order aside, arguing that the ordered timelines were not operationally feasible.

Money has continued to move while the order has not been satisfied. Since July 2016 more than 10,050,000 products, services and supports have been approved under Jordan's Principle, against roughly 10 billion Canadian dollars announced since 2016. On 2026-02-26 the department announced a further 1.55 billion Canadian dollars to extend the program to 2027-03-31, and the organisation that brought the original complaint responded that funding is necessary but does not by itself cure non compliance with a binding order.

Whose problem is this?

The party under the order is the same party that operates the program the order is about, and the body that made the order is not the body that delivers the service. Nothing in that arrangement supplies a step that happens automatically when a deadline passes.

Where does this problem end?

The boundary matters because three separate tracks run under the same 2016 ruling and are easily read as one. Two of them are about money for past harm and about how the system should be rebuilt over a decade, and this document is about neither of those. It is about the requests that were filed and are still waiting.

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

Now

The last four rows are counted on a fiscal year or cumulative basis and every row above them is counted at a calendar date. They are kept apart because they are not comparable line by line, and the increase of about 367 percent describes approvals between two fiscal years rather than any movement in the pile.

The state the orders describe is written into the orders themselves. An individual urgent request is to be answered within twelve hours and an individual non urgent request within forty eight hours, a group urgent request within forty eight hours and a group non urgent request within one week, and under the 2025 ruling the requests that missed those windows are to be cleared. Nothing in the material opened here reports a date by which the federal government accepts that this will be true, and nothing reports a single published figure for how often the ordered times are met.

How big is it?

What is counted is requests, and the count has stayed inside a narrow band for more than a year. About 140,000 were pending at 2024-12-04, about 135,000 at 2025-02-20, about 140,000 at 2025-06-04 and about 130,000 around January 2026. Between December 2024 and June 2025 about 100,000 new requests arrived, so the pile held its level while a large volume passed through it.

A pending request is not a child. One child can file several requests across a year, one group request can cover many children at once, and no source opened here gives the split between individual and group requests inside the pile or any ratio of requests to children. The direction of that correction is not even fixed, because counting requests overstates the number of children wherever one child files repeatedly and understates it wherever a single group request covers a whole community.

The population figures that exist do not compose into a denominator either. The 2021 Census recorded 747,680 people with Registered or Treaty Indian status, of whom 40.6 percent lived on reserve and 59.4 percent lived off reserve, before any adjustment for 63 reserves that were incompletely enumerated. A separate census figure puts 27.1 percent of the First Nations population under the age of 15. Multiplying the two would produce a number that looks firm and carries at least three unmeasured corrections inside it, all pointing the same way: eligibility runs to the age of majority rather than to 15, it reaches children resident in Yukon and children recognised by their own Nation while living off reserve, and registration status does not track that recognition.

Two groups sit outside the denominator for reasons of their own. Inuit children are served through a separate initiative and do not appear in this backlog at all, and Métis children have no equivalent federal initiative, which is an absence rather than a smaller share.

Needs a new measurementthe number of children. No source opened here counts the children eligible under Jordan's Principle, and none counts the children behind the requests that are pending. The population file for this document therefore records the estimate as not derivable rather than assembling one out of parts that do not fit together.

Under what conditions does it arise?

1. The party under the order operates the program the order is about. The tribunal can find non compliance and order the pile cleared. Answering the requests is work that only the department can do, and the department is the respondent. There is no third party who can step in when a deadline passes.

2. Arrivals and capacity move independently. About 100,000 new requests arrived across six months while the pile stayed near its level. Approvals rose about 367 percent between FY2021-22 and FY2023-24. A system can therefore process far more than it once did and finish the year with the same pile it started with, and both facts are true at the same time.

3. Adherence is not measured in public. The federal document that reports on this program records approvals and budget totals. It carries no figure for the pending pile and no figure for how often the ordered response times are met. Every backlog number in this document surfaced through litigation records, cross-examination or answers given to reporters, which means the measure appears only when someone compels it.

4. The order can be contested while the clock runs. The application for judicial review was filed about four weeks after the summary decision. The requests that were pending on the day of that filing remained pending, and the material opened here reports no step that suspended either the order or the arrivals while the challenge proceeds.

5. The program runs on fixed funding horizons. The most recent announcement extends it to 2027-03-31. A service whose continuation is renegotiated every few years is difficult to staff for a queue that has to be worked down rather than merely kept level.

A deadline that nobody publishes performance against is a deadline in name only, because the distance between the rule and the practice never surfaces as a number that anyone has to answer for.

What has been tried?

Two directions were tried and they run against each other. The tribunal route produced a finding, then deadlines, then a wider definition of who is eligible, then an order to clear what had accumulated. The federal route answered the last of those in court rather than in the queue, and the negotiated route that might have replaced both was voted down and then narrowed to a single province.

What was found?

Why is it still unsolved?

Enforcement absent — the rule was set nine years ago, the deadline was made binding eight years ago, the failure to meet it was ruled on, and none of that produced a step that happens without the agreement of the party that is failing to meet it.

A tribunal order that is legally binding still needs someone to carry it out, and here the only body that can carry it out is the body the order was made against. Deciding a pending request means a person reading it and a budget line answering it, and both of those sit inside the department. So the order arrives, the department receives it, and what follows next is whatever the department does next. The tribunal can rule again on that, which is how a non compliance motion came to exist in the first place, and a second ruling has the same shape as the first.

The second part is that contesting an order is a normal legal step and it costs the person waiting nothing that anyone counts. The application for judicial review was filed about four weeks after the summary decision. From outside it looks like the matter is proceeding, because it is proceeding, and the requests that were pending on the day of the filing were still pending months later. A challenge is not delay in any formal sense, but the pile does not distinguish between a request that is being argued about and a request that is being ignored.

The third part is the quietest and it is the one that holds the pattern in place. Nobody publishes how often the twelve hour deadline is met. The federal document that reports on this program reports how many approvals were issued and how much money was announced, both of which have risen sharply, and it says nothing at all about the pile or about adherence. A rising approval count and a flat backlog are perfectly compatible, and only one of the two is published on a schedule. Every figure in this document for the pile had to be extracted through litigation or through a reporter asking, which means the measure of the failure exists only when someone goes and gets it.

What observation would mean it is solved?

Candidates — a. the pending pile falls toward zero and stays there across several published dates. b. a rate of adherence to the ordered response times is published and rises. c. the federal government withdraws the judicial review application and the tribunal records compliance rather than a further breach.

a alone is weaker than it looks. A pile can fall because requests were decided, because requests were reclassified, because fewer were filed, or because what counts as pending changed. Approvals rose about 367 percent between two fiscal years while the pile did not fall, which shows that throughput and backlog move independently in this system. The pile is also visible only through litigation records and reporting, so a fall could as easily be a change in what somebody chose to count as a change in what happened to a child.

b does not exist yet, and that is the finding. No source opened here publishes any adherence rate against the twelve hour and forty eight hour orders. The observation that would settle whether the order is being obeyed is not currently measured anywhere the public can see, so before b can be read it first has to be created. That makes it the strongest of the three, because unlike a it cannot be moved by reclassification, and unlike c it is about service rather than posture.

c counts posture, not service. Withdrawing a court application would end the contest over the order. It would not by itself answer a single pending request, and a department that stopped litigating while the pile stayed at 130,000 would have changed its legal position and nothing else.

The honest reading is that b has to be built before it can be observed, a has to hold across several dates and be read against the counting basis, and c is the weakest of the three despite being the one most likely to be reported as a resolution.

What is it connected to?

Fills with researchthe relation to Indigenous child welfare failures already documented in this corpus for Australia and for New Zealand was not established in this round. The mechanisms differ, because those documents concern removal of children into state care and the funding balance between prevention and placement, while this one concerns services delivered to a child who remains at home, but relation type and evidence grade were not confirmed here. Two adjacent Canadian structures were named and not pursued: the processing backlog at the Specific Claims Tribunal, and the long term drinking water advisories on reserve. The relation between this backlog and the two other tracks running from the same 2016 ruling, compensation and long term reform, is described in the scope block above but was not evidenced as a causal link.

What these sources do not say

• There is no standing public measure of the pile. Every backlog figure in this document came out of a court file, a cross-examination or an answer given to a reporter. The federal reporting document opened here records approvals and budget totals and carries nothing on the pile and nothing on adherence to the ordered times. That is not an absence of data, because the department knows the number well enough to state it under oath. It is an absence of publication.
• Where the pile is concentrated. Nothing opened here breaks the backlog down by province or territory, or by whether the requests are for health, education or social services. The one structural split reported is between requests and appeals, at about 135,000 and 727 respectively at 2025-02-20.
• How many children are behind the requests. The individual and group split of the pile is not published anywhere opened here, and no ratio of requests to children is given, which is why the population file records the estimate as not derivable.
• How many children are eligible. No government or independent count adds up the four eligibility categories the tribunal recognised. The census figures available cover a different age boundary and a different definition, as set out in the size block.
• Whether any independent body has audited this. No audit by the Auditor General or a comparable independent office was found in this round. Every figure here is either a federal statement or something produced through the tribunal proceedings, so third party verification is unestablished rather than absent.
• What became of the judicial review. The application was filed on 2024-12-20 and the sources opened here report it as proceeding. No decision of the Federal Court on it was found.
• What answer was given to a tribunal deadline reported for 2025-08-29. Search results referred to a date by which Canada was to respond on resuming national negotiations. Every attempt to open the reporting behind it was refused, so neither the deadline nor the answer is confirmed here.
• Which of two dates in one source is right. The article carrying the 135,000 figure is dated 2026-02-27 and prints that figure as of 2025-02-20. This document keeps the as-of exactly as printed and does not resolve the discrepancy, which may be an error in the original or a re-citation of an earlier figure.
• What the Federal Court decided in Schofer v. Canada, 2025 FC 50. It is listed below as a source described as recognising incidental costs within Jordan's Principle. Nothing opened here establishes the scope of that decision or whether it changed how requests are handled, so no claim rests on it.

See the evidence

No tribunal decision and no court filing was read in full. The 2016, 2017, 2020 and 2025 rulings are described here from a federal timeline page, from the list of orders kept by the organisation that brought the complaint, and from an encyclopedia entry, which is why evidence_tier is secondary and why no wording from any order is reproduced above. The evidentiary weakness of this document is concentrated in one place and should be read plainly: the backlog series is its central fact and only one of its four points comes from a source that opened, that one carrying a publication date about a year after the as-of it prints. The other three come from reporting that refused automated retrieval, and their underlying material is a court record and sworn cross-examination rather than a published statistic. Four rows above therefore carry no URL for that reason and a fifth carries none because a federal news release refused retrieval; all five are kept rather than deleted, because a table that looks complete by being shorter is the worse outcome. Where opened sources overlap they agree: the response times of 12 and 48 hours appear in two, the 2016 finding of discrimination appears in two, and the separation of the compensation track from the reform track appears in two. The strongest evidence here is federal and points the same way as the complaint, since the department reports approvals rising about 367 percent between two fiscal years and reports nothing at all about the pile or about adherence, and that silence in an official reporting document is itself one of the findings above. This is a research-based definition, so observation_refs is empty and provenance_mode is press-derived.

People affected

Estimated range Not derivable

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