Exams: HSK and Beyond (as of 2026-09-06)

This chapter answers a practical question: which Chinese proficiency exam, if any, should a beginner be aiming at, and what does passing one actually prove? It explains the HSK (the internationally default test), the standard it is built on, the speaking test that sits beside it, and the mechanics of booking, paying, and getting a score — plus the main alternative, TOCFL. Everything exam-operational here is date-stamped, because this is a chapter about a system that is mid-replacement: as of 2026-09-06 the HSK is officially in transition, and any statement about levels, dates, or fees is a statement about that moment. Read it once now for orientation, and re-check the live facts at booking time.

What the HSK is, and how big it is

HSK stands for Hànyǔ Shuǐpíng Kǎoshì (汉语水平考试, "Chinese Proficiency Test"). It is the People's Republic of China's standardized Chinese proficiency test for non-native speakers, currently issued and administered by Chinese Testing International (CTI) within the Ministry of Education's CLEC/Hanban system. That institutional lineage is the reason HSK behaves like an official document rather than a commercial product: it is the test institutions recognize, and its administering body is the test's publisher.

The system most learners have actually taken is HSK 2.0 — the six-level ladder. Official pages report that it assesses listening, reading, writing, and speaking across six levels, with HSK 1 as the recognized beginner-entry level (attribute this to the official description rather than treating it as a fixed fact; the next section explains why the ladder itself is moving).

Scale, as reported: institutional press relaying official figures report over 819,000 foreign candidates worldwide in 2024, with cumulative participation of about 7.93 million person-times. "Person-times" means sittings, not people — one learner who takes the test three times contributes three. So the annual figure and the cumulative figure measure different things and should never be compared to each other directly (as of 2026-09-06).

You may also meet a larger 2024 number. A 1.28 million figure circulates in some learner materials; it has no resolvable source and is not printed here as fact. If you encounter it, treat it as an unverified variant whose definition is unknown — it may count a different population or a different unit, and nobody can say which.

The delivery network is reported at 168 countries and 1,477 centers, up from 34 countries and 61 centers in the early days — but that figure is hedged for two reasons. First, sources are ambiguous about whether they count test centers or individual test rooms, which can differ a lot. Second, a competing 2025 report gives 157 countries and 933 venues, and the two have not been reconciled. Directionally the network is global and large; precisely, its size is unsettled as of 2026-09-06.

For wider context, the scale anchor reported for Chinese learning overall is a cumulative international population of learners and users at over 200 million (2025 reporting). Chinese-language primaries behind it read "close to 200 million" in a Ministry of Education statement from December 2023 and "close to 210 million learning and using it abroad" in May 2026 reporting. Note the definition: this counts learners and users cumulatively, which is not the same population as HSK candidates, and not the same thing as unique individuals.

Origin dates are the softest facts in this chapter. HSK is often described as developed at Beijing Language and Culture University from 1984, designated a national-level test in 1992, and transferred to Hanban in 1999. Reports vary here — the available timeline source could not be retrieved — so treat that chronology as plausible background, not as anchors you would cite to an institution.

The standard underneath: GF 0025-2021 and "three tiers, nine levels"

Every HSK level is a claim about a position on a written standard, so the standard matters more than any single test form.

The anchor is 《国际中文教育中文水平等级标准》 (Guójì Zhōngwén Jiàoyù Zhōngwén Shuǐpíng Děngjí Biāozhǔn, "International Chinese Language Education Proficiency Standards") — commonly called HSK 3.0 — promulgated in 2021 by China's Ministry of Education and the State Language Commission. Its designation is GF 0025-2021 and it took effect on 2021-07-01, confirmed against the MOE announcement itself. Use the year of promulgation, not a day, when you date it. (You may see a different standard number cited for this document in secondary materials; that citation is unverified and should not be relied on.)

What the standard does is reorganize proficiency into sān děng jiǔ jí (三等九级, "three tiers, nine levels"): beginner, intermediate, and advanced tiers, each split into three sub-levels. This is the structure that replaces the six-level 2.0 ladder — the replacement being exactly the transition described in the next section.

The same standard's assessment architecture combines four quantitative baselines — syllable counts, character counts, vocabulary counts, and grammar items — with five skills: tīng / shuō / dú / xiě / yì (听/说/读/写/译 — listening, speaking, reading, writing, and translating). This is worth understanding even though it is not test mechanics: it is the frame that bounds what any "what's on the exam" table can claim. A test form is one operationalization of that grid, not the grid itself.

For the legacy ladder, the HSK 2.0 six-level syllabus specifies cumulative vocabulary of approximately 150 / 300 / 600 / 1200 / 2500 / 5000 words for levels 1 through 6. Always label those numbers "HSK 2.0 (legacy)" — they are not the 3.0 requirements, and no per-level character counts for 2.0 were established.

For 3.0, one directional statement is supportable and nothing more specific is. Tertiary reports say 3.0 substantially raises vocabulary demands at the upper levels — roughly double 2.0's 5,000-word ceiling. Per-level word counts for 3.0 are not yet documented in a source that holds up, so any table of "words per 3.0 level" you find, including confident-looking ones, is guesswork dressed as reference. There is also no official figure for how many hours of study a level takes: HSK 3.0 prescribes no learning hours, and the CEFR (the European scale other exams map onto) specifies none either. Hour tables you encounter are marketing-derived estimates.

CEFR banding, and what a score means

The banding you will see everywhere — levels 1–3 ≈ A1–A2, levels 4–6 ≈ B1–B2, levels 7–9 ≈ C1–C2 — appears only in tertiary academic-copy pages. Its official status and its documented limits are unverified, so use it as a rough cross-reference and not as an equivalence you can quote to an institution.

The deeper point is about what the certificate evidences. A score measures performance on exam-defined content. The popular equation "HSK 6 = fluent Chinese" overstates it: that is an interpretation people make, not a finding of the test. A high score is strong evidence you can handle the tasks the test defines, and weaker evidence about everything else.

The rollout is a transition, not a switch

This is the single most time-sensitive fact in the book: as of 2026-09-06, HSK 3.0 is transitioning, and its own rollout status is contested in the record.

Two dates must be kept separate, because collapsing them is the most common error in learner materials. The standard was promulgated in 2021. The exam trial launched years later: Chinese outlets reported on 2025-12-19 that the global trial opened on 2026-01-31, with trial registration opening 2025-12-19 — paper registration closing 2026-01-07 and computer-based 2026-01-15. These dates are tertiary-sourced but mutually consistent across reports.

Now the unresolved part. Launch-era media described the trial as opening levels 1–6, presenting that as the testable band. A report dated 2026-09-06 instead describes levels 7–9 as operational, with a first administration on 2025-12-07 — earlier than the 1–6 trial. Neither side is fully verifiable, and as of 2026-09-06 this chapter asserts neither: it does not claim 7–9 are unadministered, and it does not claim 7–9 are fully open. If you need one of those propositions to be true for your plans, go verify it against the administrator's own announcements before you act.

What is safe to say: 3.0 is being introduced while 2.0 is still in the record, and the transition began in July 2026 with no documented end date. Tertiary guides report that previously earned HSK 2.0 certificates and existing progress remain recognized during the transition. That report is reassuring but not a primary-verified guarantee, so the operational instruction is simple: when you book, check which version — 2.0 or 3.0 — is actually being offered for the level you want. Never assume 2.0 has ended; nobody in this record can establish that.

Related open questions, unresolved as of 2026-09-06: the launch date for the 7–9 band as a standing offering, and the official CEFR level-mapping table for 3.0. Also unresolved: whether the old six-level test continues in parallel through the whole pilot period, and for how long.

PRACTICAL ADVICE — reading exam news in a transition

• Decide by asking: does this source give a date and a document, or only a headline? A transition has no single switch date, so any claim of the form "HSK is now fully 3.0" or "the old test is gone" is a claim you should be able to trace.
• Keep two dates apart in your notes: when a standard was published, and when a test based on it launched. They can be years apart.
• Re-verify status facts close to your booking, and record the date you checked. This chapter's date-stamps exist because a chapter like this goes stale fast.

HSKK: the speaking test that is formally separate

HSKK — Hànyǔ Shuǐpíng Kǒuyǔ Kǎoshì (汉语水平口语考试, "Chinese Proficiency Speaking Test") — is, as of 2026-09-06, a standalone oral exam with three levels: chūjí / zhōngjí / gāojí (初级/中级/高级 — beginner, intermediate, advanced). Per official-outline-derived descriptions, it has no formal prerequisite relationship with the written HSK: neither one requires the other, and HSKK scores are reported independently. Attribute this to the official outline rather than treating it as settled mechanics, since the mechanics are moving.

Here is the finding that most surprises learners. No official, binding rule requiring HSKK in order to take HSK levels 3–6 appears in Ministry of Education publications, in the CLEIC portal, or in HSK portal publications. This is a negative-existence result from targeted searches across those domains, so its confidence is capped below established fact — but it is the reason this chapter does not present a mandatory speaking coupling as a rule.

The complication is registration behavior during the pilot. Chinese media reports from December 2025 and January 2026 describe pilot-era simultaneous registration of HSK levels 3–6 with a paired HSKK sitting, and the same pilot coverage sets certificate validity at two years. As of 2026-09-06, no official source confirms either practice. So the honest statement is: HSKK is formally independent and is not a certification prerequisite for HSK as of 2026-09-06, while unconfirmed pilot-era registration practice has bundled the two at levels 3–6. That is a description of how registration behaved at one moment, not a standing rule.

It gets messier still. Reports dated 2026-08 and 2026-09 describe co-registration for levels 4–6 — a different band from the 3–6 in the pilot-era coverage, coming from the claimed 2026 syllabus. The discrepancy is unresolved as of 2026-09-06, and no version of "HSKK is required for HSK 4–6" (or 3–6) may be published as a rule. The scope question — whether speaking co-registration is required, and for which levels — waits on the official syllabus and FAQ.

The scope of any such coupling, its effective date, and whether it is pilot-only or permanent are all unverified as of 2026-09-06, including in the 2026 policy documents of individual universities and scholarship bodies that learners often quote at each other.

Test mechanics: what you actually sit

Every paragraph in this section carries the same caveat: the official syllabus behind it has not been independently verified. The official documents behind these claims could not be retrieved, and the only source this book could reach for them is a dead tertiary page, so what follows is unconfirmed reporting pending primary verification (as of 2026-09-06) — genuinely useful for planning, not citable to an institution.

A reported August-2026 syllabus gives HSK levels 1–6 three sections — listening, reading, writing — with total test times of 55 / 65 / 80 / 105 / 120 / 130 minutes for levels 1 through 6 respectively. On the same report, level 6 adds a translation section, and translation at the upper levels is limited to five target languages: English, Japanese, Korean, Thai, and Vietnamese. Levels 7–9, per that report, use an integrated oral component and run 155–180 minutes in total. Whether the translation target language can be switched mid-test is an open question.

Scoring, as reported for 2026: each section is scored 0–100, the total is an unweighted sum of the section scores (a simple addition with no weighting applied to any part), and no official pass mark is published. Scores are reported valid for two years from the test date. On the reported rules there is no line you cross called "passing"; the number exists to be read by other people.

That distinction matters more than anything else in this section. Universities and employers publish minimum HSK scores for admission or hiring. Those are adopter rules, not exam pass marks. If a program wants level 5 at a certain score, that is that program's requirement, not a statement that lower scores "fail" — and it is also why the two-year validity clock can matter for a certificate you earned years ago.

Writing input depends on how you sit the test. Computer-based candidates type pinyin and select simplified characters from a popup grid, with no handwriting entry. Paper-based candidates handwrite — simplified characters, or traditional characters used consistently. So the practical question "can I get away with not learning to write characters by hand?" has one answer for the keyboard and a different answer for the pen; find out which you are booking before you decide what to practice.

Registration-level rules reported for the same unconfirmed syllabus: skipping levels is permitted, single-section retakes are prohibited (you retake the whole test, not the part you lost), and a same-level retake requires a 30-day cooling-off period with no limit on the number of attempts. Accommodation and disability rules for the writing section under internet-based delivery are unverified as of 2026-09-06.

Logistics: booking, money, timing, results, validity

The registration channel is not a matter of preference. www.chinesetest.cn is the sole official registration channel for HSK and HSKK worldwide; it is also the authoritative place to check current dates, formats, and results policy. Book through it.

Reported timing rules, as of 2026-09-06: paper-test registration closes roughly 27 days before the exam date, while computer-based bookings are accepted from at least 3 working days out. The 27-day figure was observed on the live site and should be re-verified when you book; the difference between the two windows is large enough that it changes what "I'll sign up later" means. "Working days" here means business days, excluding weekends and holiday periods — a different clock from calendar days.

Cadence is best stated qualitatively. The reported pattern is fixed Saturdays roughly monthly except in July and August for paper sittings, plus rolling Monday-to-Saturday computer-based slots outside Chinese holiday weeks. Treat that as the shape of the calendar, not as a count of dates — and expect the international calendar to differ from the mainland one.

Fees. The following is a China-domestic schedule; international conditions may differ, and per-region fee data was not collected for this chapter, so do not plan an overseas budget from it. Mainland-China fees are reported as nationally uniform, rising by level:

The caveat on that table is about its paper trail, and it is a real one. The schedule is dated 2025-09-01, but the page carrying it could not be retrieved and the underlying anchor is a notice from 2012. A fee ladder that still rests on a 2012-era notice may not reflect later adjustments — price levels in an economy move, and documents do not always follow. So read these yuan amounts as unverified against a current posting, and confirm the exact fee at checkout.

Results turnaround is a range that has not been reconciled, and this chapter preserves it as a range: computer-based results typically within roughly 2–3 weeks, reported as 10–15 working days, and paper-based within about 30 calendar days (as of 2026-09-06, the spread between those reporting windows is un-reconciled). Do not plan a certificate-dependent application around the fast end of that range.

Validity: the widely reported rule is that an HSK score is valid for two years from the test date. It is attributed to 《汉语水平考试管理办法》 (Hànyǔ Shuǐpíng Kǎoshì Guǎnlǐ Bànfǎ, "Administrative Measures for the Chinese Proficiency Test"), specifically to Article 22 of MOE Order No. 51 of 2021, and it is independently mirrored in mainland university admission requirements published for 2026. The Order's own text could not be retrieved, so present this as a reported rule: plan on two years, and verify before you rely on it.

Test-day rules, as reported: a printed admission ticket plus the original identification document used at registration is mandatory, and a digital 准考证 (zhǔnkǎozhèng, admission ticket) on a phone is not accepted. Rescheduling is not offered. Cancellation refunds only apply more than 15 calendar days out, minus a 20% handling fee. All of that is hedged logistics as of 2026-09-06 — re-check it, and note which ID document you registered with.

PRACTICAL ADVICE — planning around the clock, not the calendar

• Decide by asking: which delivery mode matches how I actually produce characters? If handwriting is weak, the keyboard mode removes the hand-writing problem; if you plan to write by hand, do not discover on test day that no paper sitting is available near you.
• Build in the slow end of the results range, especially if a school or employer needs the score by a deadline.
• Do the 27-day / 3-working-day difference first: paper seats close far earlier than computer-based bookings, so pick the mode before you pick the date.
• Photograph nothing as your only copy of an admission ticket; print it.

How a certificate is used, and what it unlocks

A certificate is a message you send to someone else, so its value is set by who reads it.

What holds up: HSK scores are read by institutions as evidence of Chinese ability against a published standard, which is why the same score appears in school and program requirements. What does not hold up as fact: specific current thresholds. The HSK levels commonly named in scholarship and university requirements — levels 4 and 5 depending on program — are reported but unverified as of 2026-09-06, and threshold details differ by program, by country, and by year. Likewise, no list of which authorities accept this test — or any other — is asserted here; the notice you apply under is the only place that question gets answered.

The practical consequence is that the certificate's usefulness is decided by the document you are applying under, not by the exam. Two years of validity means a score can expire before your application does. And because scores measure performance on exam-defined content, the strongest claim a certificate supports is narrow and true: you handled this test's tasks at this level on this date.

PRACTICAL ADVICE — before you register for anything

• Decide by asking: who is going to read this certificate, and what do they say they require? If nobody will read it, the exam is a study schedule you paid for — useful, but not the same thing.
• Prioritize the certifying body's own study guide, exam outline, and official practice tests over third-party brands. The official materials define the tasks; commercial books interpret them, and interpretations drift during a transition.
• Check the specific admission or scholarship notice you will apply under before choosing a level, and book through the official channel only.
• During a rollout transition, confirm which version of the test you are registering for, and save a dated copy of what the site said.

Beyond the HSK: TOCFL, YCT, and the rest

The main alternative to HSK, for learners whose target is Taiwan-based, is TOCFL — the Test of Chinese as a Foreign Language. It is Taiwan's official foreign-language Chinese proficiency test, organized by the Steering Committee for TOCFL (SC-TOP / 华测会, Huácèhuì) and aligned with the 華語文能力基準 (Huáyǔwén Nénglì Jīzhǔn, "Chinese Proficiency Standards"). Attribute its organizer and alignment to SC-TOP's own materials; the organizing body's own site could not be retrieved for this chapter.

Structurally, the standard TOCFL assesses listening and reading only — it has no speaking or writing component. That is a genuine difference from HSK, not a detail: a test that never asks you to speak or write is measuring a different slice of ability, and you should not expect it to certify the parts it does not touch.

Its level structure is disputed in the sources. Some describe three bands (A, B, C) containing seven levels; others describe four bands of two levels each, giving eight; a five-band variant also appears. So this chapter states no exact TOCFL level count as fact: describe it as a graded band system running from beginner to fluent-proficient, with CEFR-style anchors, and go to sc-top.org.tw for the current structure (as of 2026-09-06). Vocabulary scaling is reported on the 2023-revised lists as running from about 424 words at Level 1 to about 10,000 at Level 7 — a single unresolved citation, so hedge it. Its overseas availability is confirmed in outline; the reported five-year certificate validity, the current structure, and the official word lists are all unverified as of 2026-09-06. Test fees are not stated here. And how TOCFL bands map to CEFR levels and to HSK levels for a beginner's purposes is an open question, not a solved conversion.

EDITORIAL ANALYSIS — choosing between HSK and TOCFL

Frame the choice by goal and target institution, not by which test is reputedly harder. Mainland-China study or work favors HSK, because that is the instrument mainland institutions read. Taiwan-oriented contexts favor TOCFL, because that is the instrument those institutions read. Difficulty comparisons between them are not the decision variable; recognition is. This is analysis for orientation, not a sourced fact.

For young learners, the reported landscape includes YCT — a Chinese test for children, reported with a four-level structure, an age range of roughly 6 to 15, and administration by the same body as HSK. Every one of those details is reported and unverified as of 2026-09-06, with no retrievable source behind them in this record, so confirm directly with the administrator before making any plan on them.

Business-oriented Chinese testing exists as a separate family aimed at workplace proficiency rather than general proficiency. This chapter carries no verified specifics for it, so treat any details you find elsewhere as needing direct confirmation from the administering body.

PRACTICAL ADVICE — picking an exam track

• Start from the recognition side: name the institution you want to satisfy, then read which test and which score it lists.
• If the institution lists nothing, ask them, in writing, before you register. An exam taken for the wrong reason is a fee and a Saturday.
• If you are a minor's parent or teacher, verify the children's test structure with the administrator rather than from summary pages, since the reported details here are unverified.

A timeline you can actually date

Reading exam news goes much better if you keep the dates in order and labeled. This is the whole documentary record this chapter rests on, as of 2026-09-06.

Notice what is missing from that list: a date on which the six-level test stopped. Nobody in this record has one.

The two ladders, side by side

The vocabulary ladder below belongs to HSK 2.0 and is labeled as such everywhere it appears. The minutes column comes from the unconfirmed reported 2026 syllabus and describes a test form, not the standard.

Three readings of that table are worth stating flatly. The 2.0 character counts per level were never established, so a "characters per level" chart you find is not filling a gap — it is covering one. The 3.0 quantities per level are not documented, and the only supportable statement about them is directional: upper-level vocabulary demands rise substantially, roughly double the old 5,000 ceiling on tertiary reports. And the minutes are not a difficulty scale; a longer test is a test with more sections, not a harder language.

Status board as of 2026-09-06

Confusions worth untangling

Most bad exam advice is one of these pairs collapsed into the other.

• The standard is not the exam. A 2021 document defining nine levels does not itself sit in a room in 2026.
• Promulgation is not administration. The year the standard appeared and the date the trial opened are years apart.
• A reported practice is not a rule. Pilot-era registration bundling and a standing requirement are different claims.
• An adopter minimum is not a pass mark. A university's line is that university's requirement.
• Person-times are not people. Cumulative sittings count visits; candidate counts count test-takers in a year.
• Working days are not calendar days. Two weeks of working days can be most of a month.
• A range is not a promise. Results reported at 10–15 working days means plan for the far end.
• A cumulative-learners figure is not a candidate figure. Over 200 million learners and users is a different population from 819,000 examinees in 2024.
• "Levels 1–6 is the trial band" and "levels 7–9 are operational" are the two sides of an unresolved conflict, not a before-and-after story.

Things this chapter will not let you conclude

• That HSK 2.0 has ended. Nothing here establishes it, and as of 2026-09-06 nobody in the record can.
• That levels 7–9 are open for booking, or that they have never been administered. Both propositions are asserted by some report; neither is verified.
• That 1.28 million foreigners took the test in 2024. That figure is an unverified variant with an unknown definition.
• That any specific number of words or characters belongs to a 3.0 level. Only the direction of change is supportable.
• That a particular score is "passing". No official pass mark is published in this record.
• That you must register for HSKK to sit a written level. Reports say it happened in the pilot; no binding rule is confirmed as of 2026-09-06.
• That the mainland fee ladder is what you will pay, or that the amounts are current. The anchoring notice is from 2012.
• That TOCFL has a fixed number of levels. Sources disagree, and the chapter states no count.
• That hours of study buy a level. HSK 3.0 prescribes no learning hours and the CEFR specifies none; hour tables are commercial estimates.
• That the 1984/1992/1999 origin chronology is settled. Reports vary.

How to check a claim about these exams in one sitting

1. Identify which object the claim is about: the standard, the test form, the registration rules, or an institution's requirement. Claims rarely transfer between them.
2. Look for a date on the claim itself. If a status claim has no date, it is not usable in a transition.
3. Ask whether the source is the administrator's own publication or a copy of one. Tertiary copies are how the conflicting 1–6 and 7–9 stories both reached print.
4. Check the offering you can actually book — chinesetest.cn for HSK and HSKK, sc-top.org.tw for TOCFL — rather than an article about it.
5. Save what you saw, with the date. During a transition your own dated screenshot is the best evidence you will have later.
6. For anything an institution requires, use the notice you are applying under. That document, not this chapter, controls your outcome.

PRACTICAL ADVICE — if you will sit only one exam

Pick the exam your target reader recognizes, then pick the level you can evidence rather than the one that sounds impressive. Study from the official guide, outline, and practice tests, because those define the tasks and third-party materials interpret them during a transition. Choose the delivery mode by your weakest skill, since keyboard and paper modes ask you to produce characters differently. Check the version being offered, note the close date for that mode, and let the results range — not the fastest reported turnaround — set your deadline. Then let the certificate's validity clock start where your application calendar can still use it.

What changes in the next year, and what does not

The parts of this chapter that will go stale are the parts with dates attached: which levels are open, what the trial band turned out to be, whether the two competing network counts were reconciled, whether the 2012-anchored fee ladder still matches what the checkout page says, and which level band — if any — ends up bundled with the speaking test. The parts that will not change are the conceptual ones.

A test still measures the content it defines, so no score ever meant more than the tasks behind it. An exam board's schedule still answers to a school's admission notice, so the document you apply under still outranks any summary. Person-times still count sittings, so scale claims still need their definition before they need their number. And a standard published in 2021 still takes years to become something you can book — which is the single most useful thing to remember the next time a headline claims a system changed overnight.

Where the confidence sits in this chapter

Not every sentence here carries the same weight, and the differences are deliberate.

Stated flatly, as established: the standard's designation and effective date; its three-tier, nine-level organization; the four quantitative baselines across five skills.

Attributed, because the source is official description rather than independent verification: that the legacy test assesses four skills across six levels; the legacy cumulative vocabulary ladder, carried here as the 2.0 ladder as published in the standard's own secondary coverage; that HSKK is a standalone oral exam with three levels; that TOCFL is organized by SC-TOP and aligned to its own published standard.

Hedged estimates, where figures exist but disagree or rest on one report: the candidate counts and cumulative sittings; the test-network size; the direction of 3.0's upper-level vocabulary increase; the two-year validity rule; the reported fee ladder and cadence; TOCFL's vocabulary scaling.

Unconfirmed reports, capped by the absence of any retrievable primary document: every section count, timing, score-band, retake, and writing-input detail in the mechanics section. As of 2026-09-06 the official syllabus behind them has not been independently verified.

Interpretation, never test fact: that scores do not equal fluency; that the CEFR banding is a rough cross-reference.

Advice, labeled as such and carrying no evidence: how to choose a test, how to read requirement language, what to prioritize in preparation.

Terms this chapter uses

• Person-times — a count of sittings, not of people. One person testing three times contributes three.
• Working days — business days, excluding weekends and holiday closures; a longer window than the same number of calendar days.
• Cumulative vocabulary — the total a level expects you to know, which already includes every lower level's words. Adding the per-level figures double-counts.
• Unweighted sum — a total produced by simple addition, with no section treated as more important than another.
• Pass mark — the score at which a test declares success. None is published in this record; do not speak as though one exists.
• Adopter minimum — a score another institution requires, which is that institution's rule and not the exam's verdict.
• Band / tier — a named grouping of proficiency. Tiers are the standard's three top divisions; bands are the graded steps used by TOCFL.
• Banding — a claimed correspondence between two scales, here HSK levels and CEFR letters.
• Primary or official publication — a document from the body that runs the test, as opposed to a copy of one.
• Tertiary source — a report or guide describing official material. Several of this chapter's dates are tertiary but mutually consistent.
• Retrieved — a source this book could actually open and check. When a source could not be retrieved, the fact resting on it stays hedged.
• Negative-existence finding — a search result saying no such rule was found in the places it should be. Strong, but capped below established fact.
• Transitioning — a system mid-replacement, where two versions coexist and no end date has been documented.

Questions this chapter leaves open

Framed as unresolved rather than answered, as of 2026-09-06:

• When, if ever, levels 7–9 become a standing offering rather than a contested one — and what the official correspondence between this scale and the CEFR turns out to be.
• Whether the six-level test continues in parallel for the whole pilot period, and for how long.
• Whether a candidate may switch translation target language partway through an upper-level paper.
• Whether any official readiness benchmark exists at all — a stated study duration, or a word count that counts as ready. None is evidenced here.
• How TOCFL's bands relate to the CEFR, and to HSK levels, in a way a beginner could rely on.