Canada's immigration law requires detention reviews within 48 hours, within 7 days and then every 30 days but sets no maximum length — 163 of the 3,598 people CBSA detained in 2025 were held more than 99 days

What is happening?

Canadian immigration law lets the federal government detain people for immigration purposes and requires that the detention be reviewed on a fixed schedule, but it sets no maximum length of detention. Section 57 of the Immigration and Refugee Protection Act requires the Immigration Division to review the reasons for detention within 48 hours or without delay, again within the following 7 days, and then at least once every 30 days. The section governs how often detention is reviewed, not how long it may last.

The Library of Parliament, in a HillNotes research summary published 2025-03-21, states plainly that there is no limit to the period for which a person can be held in immigration detention.

Most detentions are short. According to the Canada Border Services Agency statistics page, 3,598 people were detained in 2025, with an average stay of 18 days and a median of 5 days, and 1,321 of them were held 24 hours or less. The same page counts 163 people held more than 99 days in 2025. In the second quarter of 2026, April to June, 1,382 people were detained and 51 were held more than 99 days. The absence of a ceiling matters for that long tail, and the only built-in instrument that bears on it is the next 30-day review.

On 2026-03-23 the UN Human Rights Committee, in its concluding observations on Canada, recommended that Canada establish a statutory time limit on the duration of immigration detention, with a follow-up reporting deadline of 2029-03-19.

Whose problem is this?

The body that would have to set a limit is Parliament, while the bodies that apply the current regime work inside a schedule of reviews that never reaches a point where detention must end.

Where does this problem end?

Individual detained people and their families are not described here; the document speaks only in aggregate counts and in the wording of the law.

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

Now

The fiscal year figures from the Library of Parliament are kept out of this table because they rest on a different period: for 2023 to 2024 the average detention was 19 days and the median 3 days, and from 2012 to 2013 through 2023 to 2024 half of all detained people were held 3 days or less.

Needs a new measurementthe target state: no Canadian source opened here sets a maximum length of immigration detention or an official target for how long detention should last. The CBSA describes detention as a last resort and says alternatives are considered first, which is a principle rather than a duration. The one stated benchmark is external — the UN Human Rights Committee recommendation of 2026-03-23 to establish a statutory time limit — and it is recorded here as a treaty body recommendation, not as a Canadian target.

How big is it?

Between 163 and 3,598 people in 2025. The lower bound counts the people the CBSA reports as held more than 99 days in 2025, the group for whom the missing ceiling has the clearest effect. The upper bound counts everyone detained that year, each of whom was held under a regime with no maximum.

The two numbers answer different questions. The first says how many people actually reached a long detention, and the second says how many were exposed to a rule that never forces detention to end.

Neither counts how long the longest detentions ran. The CBSA page opened here shows only a top band of more than 99 days, so a person held four months and a person held two years fall into the same cell.

For minors, the Library of Parliament reports 6 detained in 2023 to 2024, all unaccompanied; the CBSA 2025 figure for minors is withheld on its page.

Under what conditions does it arise?

1. Review replaces a limit. The Act asks the Immigration Division, every 30 days, whether the grounds for detention still exist. It does not ask whether the total time held has become too long by a fixed measure, so the same detention can pass review after review.

2. The courts kept the regime and left a safety valve rather than a clock. On 2020-08-07 the Federal Court of Appeal, in 2020 FCA 130, held the regime without a time limit constitutional. As reported in commentary on the decision, continued detention must keep a link to an immigration purpose, and release can be warranted under the Charter where the length or conditions of detention seriously infringe liberty. The argument for a fixed delay ceiling modelled on the criminal trial framework was not accepted.

3. The long tail is small next to the average. With a median of 5 days in 2025, the overall picture looks brief, and the 163 people held over 99 days are a small share of the total. A problem concentrated in a few hundred cases a year does not show in the headline average that public reporting leads with.

4. The place of detention has shifted, not the duration rule. Provinces ended the use of their jails for immigration detention, the last by 2025-09-15, and from 2025-07 the CBSA began using a designated area of a federal penitentiary. The question of where people are held moved; the question of how long they may be held did not.

What has been tried?

The attempt that changed something changed the place of detention, and the one that asked for a limit came from outside the country and has a reporting date three years out.

What was found?

Fiscal year findings from the Library of Parliament, opened directly and graded medium because they summarise agency data rather than publish it: an average of 19 days and a median of 3 days in 2023 to 2024, and 6 minors detained that year, all unaccompanied.

Why is it still unsolved?

Institutional gap — the law builds in a schedule for asking whether detention should continue, but no point at which it must stop.

A review cycle and a ceiling do different work. A review asks each month whether the grounds for detention still exist, most often that the person is unlikely to appear, and if the answer is yes the detention continues for another cycle. A ceiling would ask a different question that the current law never puts: whether the total time has passed a limit regardless of the grounds. Because only the first question exists, a detention can run for any length provided each review finds the grounds intact.

The 2020 appellate decision closed the most direct route to change through the courts. By finding the regime constitutional while leaving release open in extreme cases, it placed the control on a case by case Charter assessment rather than on a fixed number of days. That leaves the length question with Parliament, and the sources opened here record no completed legislative step toward a limit.

The pattern is also hard to see. With a median of a few days, public figures read as a system of short detentions, and the people held for months appear only as a single top band of more than 99 days. A problem that sits in a few hundred cases each year, inside a band with no upper edge, produces no average that would draw attention to it.

What observation would mean it is solved?

Candidates — (a) Parliament amends the Act to set a maximum length of immigration detention, in force (b) the CBSA count of people held more than 99 days falls and stays low across several years (c) Canada reports to the UN Human Rights Committee by 2029-03-19 that a statutory limit is in force.

(a) alone is weaker than it looks. A limit can be set so high, or with exceptions so broad, that the long tail does not move, and an exception for people whose identity is unconfirmed or whose removal is pending could keep most long detentions outside it.

(b) alone is worse. The count of long detentions moves with the number of arrests, removal capacity and the caseload of the Immigration Division, independently of any rule. It also cannot show what happens above 99 days, because the published band has no upper edge.

(c) alone counts paper. A report to a treaty body describes the law as the government reads it. The three have to be read together, and (b) has to be published with bands above 99 days before it can say whether the longest detentions ended.

What is it connected to?

What these sources do not say

• How long the longest detentions run. The CBSA page opened here shows a top band of more than 99 days only. Whether the page carries longer bands, such as over 180 days or over a year, and what the single longest detention was, are not shown in what was opened.
• Whether 2025 is a calendar or fiscal year. The CBSA page labels the figures 2025 full year and does not say which.
• Who is in the long tail. No source opened here breaks down the people held more than 99 days by ground for detention or by country of origin.
• How many people are held in the federal correctional facility. The sources describe the arrangement and its start date but give no count.
• Whether any bill to set a limit is before Parliament. The Library of Parliament summary says that efforts to set a limit have not succeeded. No bill text was opened, so no particular bill is described here.
• Why the government has not set a limit. No source opened here carries an official reason for keeping the regime without a maximum. The government response to the 2026 UN recommendation is not in the summary that was opened, and no other public statement of the government position on a time limit was found.
• The 2020 appellate decision itself. Its text returned an automated block, so its holding is taken from commentary. Whether a further appeal followed is not stated in the sources opened.
• The CBSA position, as published. The agency says detention is a last resort and that alternatives are considered first, and a 2023 government note places about 98 percent of people under supervision in alternatives. Those statements concern whether to detain; neither addresses how long detention may last.

See the evidence

The primary documents opened directly are the Act, the two CBSA pages, the Library of Parliament summary and the government Question Period note. The appellate decision and the UN concluding observations were not opened in their own text; both are taken from secondary sources, and the decision text and one news report on it could not be retrieved. Where sources overlap they agree: the review schedule appears identically in the Act and on the CBSA page, and the absence of a limit is stated in the Library of Parliament summary and follows from the Act. Fiscal year figures from the Library of Parliament and the CBSA figures labelled 2025 are kept apart throughout because their periods differ and the CBSA label does not say which year type it uses. This is a Path A output (research-based definition), so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 163–3,598 As of 2025, full year as labelled by CBSA

Sensitivity The interval runs from the people who actually reached a long detention to everyone exposed to a regime with no ceiling. The low end cannot show how far beyond 99 days any detention ran, because the published band has no upper edge. The page does not state whether people detained before 2025 and still held during it are included, and it does not say whether 2025 is a calendar or fiscal year. Fiscal 2023 to 2024 figures from the Library of Parliament are not mixed into this interval.

Regional breakdown No source opened gives detention counts by province or region; the agency publishes national figures only, and apportioning them would be inference rather than measurement