IRCC data commentaryATIP 1A-2025-08687
中文GitHub
OPINION · Written from ATIP 1A-2025-08687 data

Canada’s most-suspected applicants are also the ones it approves most

In IRCC’s security screening, Chinese nationals account for more than a third of all screening volume — more than the next four countries (Iran, India, Ukraine, Pakistan) combined. A Chinese applicant’s chance of being sent for comprehensive security screening is more than ten times an Indian applicant’s. Yet set that screening against outcomes and the answer is awkward: China has the highest approval rate of any large source country Canada sees, 80.1%, well above the national average of 59.6% and above India’s 63.6%. Using the same ATIP file, this piece places China and India side by side and asks three things: whether extra screening is the exception or the norm for a Chinese applicant; whether the basis for that screening holds up; and what the resources it consumes actually buy.

Chinese applicants’ screening rate
3.5%11× the Indian rate, 4.1× the national average
China’s share of national screening
37%while China is only 9.0% of all applications
China’s unfavourable-result rate
0.62%below the national 0.80% and India’s 1.3%
Part one

Extra screening is the norm for Chinese applicants

Before arguing about fairness, one basic fact has to be settled: how common is it, really, for a Chinese applicant to be singled out for comprehensive security screening. China and India are Canada’s two largest immigration source countries, and setting them side by side is far more telling than comparing China against any small country — both run to hundreds of thousands of applications, so the proportions are not distorted by a tiny denominator. The two charts below give the answer: for Indian applicants, hitting a security screening is basically the exception; for Chinese applicants, it has become close to routine.

Three streams, one direction

Canadian applications run through three main streams: immigration (permanent residence), temporary-residence visas, and study permits. This chart lays out the screening rate for China, India and the national average on each stream. All three streams point the same way, and the widest gap is on the immigration stream — the one that leads to permanent status, is screened hardest, and waits longest.

Referral to comprehensive security screening — by application stream

One bar each for China, India and the national average within every stream; the length is the share of that stream's applications referred to comprehensive security screening. The bar tip shows the rate and how many times the national average it is. All three streams share one scale, so the bars compare directly across streams.

ChinaIndiaNational average
申请通道
0.0%3.0%6.0%9.0%12.0%
All applicationsPR + study permits + temporary-residence visas
🇨🇳 China
3.5%4.1× vs national
🇮🇳 India
0.33%0.4× vs national
National average
0.86%
Temporary-residence visasVisitor, study, work
🇨🇳 China
3.1%5.0× vs national
🇮🇳 India
0.26%0.4× vs national
National average
0.62%
Permanent-residence applicationsImmigration applications
🇨🇳 China
12%4.0× vs national
🇮🇳 India
1.3%0.4× vs national
National average
3.0%
Applicants referred to CBSA/CSIS for comprehensive security screening, counted across all screening types (Security / Org Crime / HIRV on the PR stream; VIT 34/35/37 on the TR stream). Denominators: permanent-residence applications received in 2025 for PR, temporary-residence visa applications received in 2025 for TR, and "All applications" is the two plus finalized study permits — matching the main report. Study-permit applicants are already contained within the temporary-residence visas, so they are not shown as a separate stream.

For every 100 Chinese who apply to immigrate, 12 are sent for security screening

Turning the rate into heads makes it more concrete. This chart is the immigration (permanent-residence) stream: each cell is one applicant, and the filled cells are the ones sent for comprehensive security screening. China’s grid has 12 cells filled, India only 1, the national average 3. To be clear, this is not "12% are refused" but "12% are, before any decision, pulled aside for an extra security investigation" — a step that usually means waiting several more months to several more years.

Out of every 100 permanent-residence applicants, how many are sent for comprehensive screening

Each grid holds 100 cells; one cell is one person who filed a permanent-residence application in 2025. The filled cells are the share referred to CBSA/CSIS for comprehensive screening, rounded to the nearest whole cell; the exact rate sits above each grid and the raw counts below.

🇨🇳 China12%
≈ 1 in 8
2,444 / 20,430
🇮🇳 India1.3%
≈ 1 in 77
1,150 / 88,395
National average3.0%
≈ 1 in 33
14,587 / 485,320
The numerator is the number of comprehensive security screenings opened on the PR stream in 2025 (Security / Org Crime / HIRV combined); the denominator is the permanent-residence applications received in 2025. An applicant opened to more than one screening type is counted more than once, so the figure is "screenings per 100 applications", slightly above the share of applicants screened.
Part two

Does the basis for the screening hold up?

A high screening rate is not in itself unfair — if it is high where there is genuine cause, that is exactly what security screening is for. So the real question is: in whose name are these screenings opened, on whom do they fall, and is it a fluke of one year. The file’s own categories happen to answer all three, and the answers point the same way: screening of Chinese nationals is concentrated on a single legal ground, falls overwhelmingly on the most ordinary visitors, and has held for six years running.

One clause of law carries 99.9% of the screening

Canada’s Immigration and Refugee Protection Act has three main grounds that can trigger comprehensive security screening: s.34 (security, covering espionage, subversion and terrorism), s.35 (human-rights violations, covering war crimes and sanctioned officials), and s.37 (organized crime). The file assigns every screening to one of them. China’s distribution is barely a distribution at all: of 15,584 screenings, 15,576 are opened under s.34. India’s makeup is clearly different, with more than a quarter under s.35. The second set of bars below also shows who these screenings fall on: of China’s screened temporary-residence applications, nearly nine in ten are ordinary visitor visas — not students, not workers.

Which legal ground each screening was opened under, and which application type it fell on

The upper bar splits the country's entire 2025 screening total by legal ground; the lower bar looks only at the temporary-residence stream, split by application type. Each bar fills its full width by share, so a very small segment is barely visible — read the exact counts in the breakdown beneath.

Security (IRPA s.34)Human-rights violations (s.35)Organized crime (s.37)Visitor visaStudy permitWork permitIn-Canada extension
🇨🇳 People's Republic of China
按法律依据
Security (IRPA s.34)
100%15,576
Human-rights violations (s.35)
0.019%3
Organized crime (s.37)
0.032%5
临时居民审查落在哪类申请上
Visitor visa
88%11,585
Study permit
6.2%815
Work permit
5.6%736
In-Canada extension
0.030%4
🇮🇳 India
按法律依据
Security (IRPA s.34)
74%3,211
Human-rights violations (s.35)
26%1,127
Organized crime (s.37)
0.12%5
临时居民审查落在哪类申请上
Visitor visa
98%3,141
Study permit
0.38%12
Work permit
1.3%40
The mapping to legal grounds comes from the file's own categories: Security Screening (PR) and VIT 34 (TR) both fall under s.34, HIRV Screening and VIT 35 under s.35, and Org Crime Screening and VIT 37 under s.37. Application type applies only to the temporary-residence stream, where “study permit” and “work permit” merge new applications with extensions.

This is not one anomalous year

Look only at 2025 and you could still call it a one-year blip. But the file covers 2019 to 2025. Add up six-plus years of screenings and China still stands alone: of 310,876 screenings, 96,574 are of Chinese nationals — 31.1%, nearly double second-place Iran and nearly four times India. In other words this is not a policy spasm but a default setting that has run steadily for six years.

Cumulative comprehensive screenings 2019–2025 — top 12 countries

The bar length is the total comprehensive security screenings opened against that country between January 2019 and December 2025; the tip shows the count and its share of the global cumulative total. China is marked warm.

国家/地区
032k64k97k
🇨🇳 People's Republic of China
96,57431%
🇮🇷 Iran
49,98816%
🇮🇳 India
24,7578.0%
🇺🇦 Ukraine
13,9544.5%
🇵🇰 Pakistan
13,2074.2%
🇸🇾 Syria
7,8702.5%
🇦🇫 Afghanistan
7,8062.5%
🇱🇧 Lebanon
7,4222.4%
🇵🇸 Palestinian Authority (Gaza/West Bank)
7,1202.3%
🇷🇺 Russia
6,7582.2%
🇹🇳 Tunisia
5,8091.9%
🇳🇬 Nigeria
5,4761.8%
The cumulative counts come from the file's "total" column (screenings opened between 1 January 2019 and 31 December 2025), combining the PR and TR streams and all screening types — 310,876 screenings in all, across 185 countries. An applicant opened to more than one screening type is counted more than once. Switch to the table view for the full list.
Part three

What do these resources buy?

Security screening is not free. Every comprehensive screening consumes CBSA and CSIS analytical labour and adds months, sometimes years, to an applicant’s wait. So the last question matters most: what does this investment buy? The two charts below read IRCC’s published screening outcomes directly, then set them against who ends up approved.

The extra screening of Chinese applicants finds almost nothing

Across every nationality a security screening almost never ends in an unfavourable result: the national rate is 0.80%. China’s is lower still at 0.62%, below both the national average and India’s 1.35%. The countries where screening genuinely catches something are Afghanistan, Ukraine and Russia, all around 3%, four to five times China’s rate. Iran, one of the two most-screened countries of all, is at just 0.20%. One caveat runs the other way: China is first in the raw count of unfavourable results, 845 of them — but only because it is screened so much. Its share of unfavourable results, 28.5%, is roughly its share of screening, 31.1%. The disproportion is in the rate, not the share — a Chinese applicant is screened at some ten times an Indian applicant’s rate, and each of those screenings is less likely to return an unfavourable result.

How often a screening ends in an unfavourable result — by nationality

Each bar is one country, filled to its unfavourable-result rate — non-favourable results in 2020–2025 over referrals in 2019–2024, offset by a year because a screening concludes about a year after referral. The dashed line is the national mean. Only countries with at least 3,000 screenings appear (14 in all), sorted so the nationalities screening genuinely catches sit on top.

Country
0%1%2%3%4%
national avg 0.80%
🇦🇫 Afghanistan
3.1%
🇺🇦 Ukraine
3.0%
🇷🇺 Russia
2.8%
🇮🇳 India
1.3%
🇱🇧 Lebanon
1.0%
🇮🇶 Iraq
0.90%
🇳🇬 Nigeria
0.89%
🇵🇸 Palestinian Authority (Gaza/West Bank)
0.83%
🇨🇳 China
0.62%
🇪🇬 Egypt
0.51%
🇸🇾 Syria
0.42%
🇵🇰 Pakistan
0.30%
🇮🇷 Iran
0.20%
🇹🇳 Tunisia
0%
Unfavourable results come from ATIP release OPP-DART-2025-34337 (non-favourable screening results by nationality and year, January 2019 – July 2025). The referral counts come from 1A-2025-08687. A suppressed cell (printed “--”, meaning a count of 1–4) is counted as zero, so every rate is a lower bound. The most recent referral cohorts have not finished concluding, so heavy recent screeners read slightly low. The national mean is 0.80%. Switch to the table view for every country.

The most heavily screened large country is also the most approved

Finally, put both facts on one chart: the x-axis is the screening rate, the y-axis the eventual approval rate, and the dot size is application volume. If screening were effectively identifying risk, the most-screened countries should sit in the bottom-right — heavily screened, low approval. China is not there. It is in the top-right: a screening rate four times the national average, yet the highest approval rate of any large source country at 80.1%. This single point all but refutes the claim that screening by nationality is filtering for risk: the population this process suspects again and again is precisely the population it ends up approving most.

Screening rate against temporary-residence approval rate — large source countries

One dot per country. The x-axis is the 2025 security-screening rate (log scale), the y-axis the 2025 temporary-residence approval rate, and the dot size tracks application volume. The two dashed lines are the national average screening rate and the national average approval rate, splitting the frame into quadrants: bottom-right is "heavily screened, low approval", top-right "heavily screened, high approval". China and India are marked warm and labelled. Only countries with at least 25,000 finalized temporary-residence applications are shown (31 in all), to avoid small-denominator artefacts.

China / IndiaOther large source countries
SCREENING RATE (LOG)TR APPROVAL RATE0%25%50%75%100%0.010%0.10%1.0%10%IndiaPeople's Republic of ChinaNigeriaPhilippinesMexicoPakistanUkraineIranNational avg screening rateNational avg approval rate
The screening rate is that country's 2025 screenings of all types divided by its 2025 applications; the approval rate is IRCC's published 2025 temporary-residence approvals divided by finalized applications. Countries with zero screenings cannot be placed on the log x-axis and are omitted. The two rates are not measured on exactly the same base: the screening-rate denominator includes permanent-residence applications while the approval rate covers temporary residence only, so read the chart as a trend rather than a precise causal chain.

Opinion: writing suspicion into the process is not the same as controlling risk

This file cannot prove any applicant innocent or suspect, nor pinpoint where IRCC’s judgement went wrong on any single case. But it can tell us where this machine’s attention goes, and what that attention ultimately produces.

Put the three parts’ answers together and one thread is clear. First, for a Chinese applicant, being sent for comprehensive security screening is no longer the exception: overall it is more than ten times the rate for Indian applicants, and on the immigration stream it is close to one in every eight. Second, almost all of these screenings are opened under the same legal ground — 15,576 of 15,584 invoke s.34 (security) — and nearly three quarters of them (11,585) land on the most ordinary visitor-visa applications, not students, not workers. A screening regime genuinely opened on case-by-case risk could hardly reach so uniform a conclusion across a population of hundreds of thousands. Third, we no longer have to infer the payoff. IRCC has since released the screening results, and they are meagre: only 0.62% of the screenings of Chinese applicants end in a non-favourable finding, below both the national average of 0.80% and India’s 1.35%, while the nationalities screening genuinely catches — Afghanistan, Ukraine, Russia — return an unfavourable result at four to five times China’s rate.

When the most-screened group is also the one its own screening clears most reliably, what the screening sorts for is not risk but nationality.

The most plausible explanation is not conspiracy but inertia: once a nationality criterion is written into the referral rules, it triggers at near-zero marginal cost, over and over, unchanged for six years. The cost does not show up on a budget line — it falls on individual applicants who wait an extra 8 months, 18 months, sometimes longer, on admission letters that expire in the waiting, on weddings and funerals missed. It also falls on the limited analytical capacity of CBSA and CSIS: hours spent on 15,584 files that will be approved eight times out of ten are hours not spent elsewhere.

Testing whether a security-screening regime works does not require seeing anything classified. It takes one question: if the extra screening tied to a given nationality were removed, would more people be stopped, or the same number? For the year this piece first asked it, the answer had to be inferred. It no longer does: IRCC has published screening conclusions by nationality, and for Chinese applicants the answer is close to “the same number” — more than 99 screenings in 100 conclude favourably. Once the extra scrutiny changes almost nothing about who is found inadmissible, the label “security screening” becomes hard to tell apart from “sorting by passport”.

Three objections to this piece, and my replies

“Screening deters on its own; genuinely high-risk people simply never apply.” This cannot be falsified or confirmed with the file. But deterrence only works if applicants know in advance that they will be screened — and IRCC’s referral rules are never published, so most applicants only discover mid-wait that they were screened at all. A mechanism you learn about only after the fact can hardly deter beforehand.

“Section 34 covers state actors, and China’s circumstances make its scope genuinely wider.” This is the objection the piece takes most seriously. But if that were the basis, the screening should concentrate on applicants connected to state institutions, not on 11,585 ordinary visitor-visa applications. The right move is to write the criterion down and publish it, not to use an entire nationality as a proxy.

“A high approval rate is exactly what proves the screening works — it clears the innocent.” If so, the number that settles it is the screening’s marginal effect: of the people screened, how many were found inadmissible. That number has now been released, and it is small — a 0.62% non-favourable rate on Chinese applicants, below the national average. An 80.1% approval rate paired with a 0.62% unfavourable-result rate reads less like a filter that clears the innocent than like a checkpoint that, for the great majority, only delays.

Method and limits

The referral file counts the times screening was opened, not the number of people screened — an applicant opened to more than one screening type is counted more than once, so a rate over “applications” should be read as “screenings per application”. The unfavourable-result rates draw on a second release (OPP-DART-2025-34337) that reports non-favourable results by nationality and year. Because a screening concludes about a year after referral, the headline rate offsets the windows by a year — non-favourable results in 2020–2025 over referrals in 2019–2024. Suppressed result counts (a printed “--”, meaning 1–4) are floored to zero, so every rate is a lower bound, and the most recent referral cohorts have not finished concluding, so heavy recent screeners read slightly low. Both make China’s low rate conservative rather than flattering. The original files are scanned images, so the figures were read by OCR and checked against the source one by one. Every chart can be switched to its underlying data table.

Read the full data report