HSK to CEFR: What's the Honest Mapping?

Search “HSK 4 CEFR level” and you will get three confident, contradictory answers in the first three results. One site says B1. Another says A2. A third hedges with “A2/B1” and moves on without explaining why it can’t commit. None of them are lying to you. They are all citing a real source. The sources just disagree, sometimes by a full CEFR level, sometimes by two.

That disagreement is not a gap in anyone’s research. It is the honest state of the field. China’s own test board and the European associations who actually teach with the test have been arguing about this since 2010, and the argument has never been resolved, it has just been quietly worked around. This article is the working-around: what each side actually claims, why a European framework was never going to map cleanly onto Chinese in the first place, and a range-based table you can use instead of pretending there is a single right answer.

Why does every HSK-CEFR chart say something different?

Because there is no official cross-walk. The Common European Framework of Reference for Languages, CEFR, was built and is maintained by the Council of Europe, for European languages. Hanban, the Chinese government body that ran the HSK until it was folded into the Center for Language Education and Cooperation, made its own claim about where HSK levels sit on that scale. European Chinese-teaching associations checked that claim against their own students and rejected it. Everyone downstream, course platforms, YouTube explainers, university placement offices, has been picking a side, splitting the difference, or copying whichever chart they found first ever since.

The 2010 claim, and why German and French teachers rejected it

In 2010, Hanban stated that the six levels of the (old) HSK corresponded directly to the six levels of CEFR: HSK 1 to A1, HSK 2 to A2, HSK 3 to B1, HSK 4 to B2, HSK 5 to C1, HSK 6 to C2. Clean, tidy, one-to-one.

The Fachverband Chinesisch, the German association of Chinese-language teachers, published a formal response the same year rejecting it. Their assessment, based on what the test actually required at each level rather than what the marketing claimed, put the real correspondence about two CEFR levels lower: HSK 3 nearer A1, HSK 4 nearer A2, HSK 5 nearer B1, HSK 6 nearer B2, not C2. Their core argument was concrete rather than rhetorical: the old HSK 5 required a vocabulary of roughly 2,500 words, thin for a level Hanban was calling advanced by European standards. French associations of Chinese teachers reached a similar conclusion independently, placing HSK 6 somewhere in the B2 to C1 range rather than at C2.

A two-level gap is not a rounding error. It is the difference between “can handle a routine transaction” and “can follow a technical discussion in your own field.” Both sides published their reasoning, and sixteen years later, neither side has walked it back.

Why a European framework was never going to map cleanly onto Chinese

Part of the disagreement is political, but part of it is structural, and the structural part is more interesting. CEFR was deliberately written to be language-independent. Its descriptors describe tasks, not vocabulary counts: an A1 speaker “can understand and use familiar everyday expressions,” a B1 speaker “can understand the main points of clear standard input on familiar matters,” a C1 speaker “can understand a wide range of demanding, longer texts and recognise implicit meaning.” Nowhere does CEFR say how many words or characters it takes to do any of that, on purpose, so the same yardstick can apply to Finnish, Portuguese, and Chinese alike.

That neutrality is also the source of the problem. The framework was calibrated against alphabetic European languages, where a learner who can hold a B1 conversation has usually also picked up basic literacy more or less for free, since the writing system just spells the sounds they already know. Chinese does not work that way. Literacy is a separate, much larger project, layered on top of spoken competence rather than riding along with it. The U.S. Foreign Service Institute’s own hours-to-competency estimates, covered in our piece on how long it actually takes to learn Mandarin, put Mandarin at roughly four times the study hours of Spanish for the same functional level, and most of that gap is the character system, not the grammar or the speaking. A can-do descriptor calibrated on French or German simply does not carry the same character-literacy assumptions when you swap in Chinese, which is exactly why a European civil servant who is B1 in Spanish and a European civil servant who is “B1” in Mandarin by HSK count are not standing in comparable places.

A framework built to be language-neutral still assumes an alphabet. Swap in a character-based language and the neutrality quietly breaks.

Merry Mandarin

HSK 3.0: the quiet correction rolling out this month

Here is the detail most articles about this topic miss because it only became true recently: the six-level HSK, the one every 2010-era CEFR chart is arguing about, is being replaced. The new standard, HSK 3.0, restructures the test into nine levels across three stages, elementary (1 to 3), intermediate (4 to 6), and advanced (7 to 9). The underlying grading standard was published back in 2021, the finalized vocabulary syllabus was only released in November 2025, the first global pilot exams ran in January 2026, and full worldwide rollout of HSK 3.0 is scheduled for July 2026, this month, alongside our own publication of this article.

The redesign is itself a tacit admission that the old top level was oversold. Under the six-level system, HSK 6 was the ceiling, and Hanban called that ceiling C2. Under HSK 3.0, HSK 6 sits in the middle of the intermediate band, with three entirely new advanced levels, HSK 7, 8, and 9, built specifically to cover the C1-C2 range the old test never actually reached. You do not add three new levels above your top level unless you have quietly accepted that your top level was not where you said it was.

By the new standard’s own cumulative vocabulary counts, HSK 1 requires around 500 words, HSK 4 around 3,245 words, HSK 6 around 5,456 words, and the combined HSK 7-9 band around 11,092 words plus 572 grammar points. That last number is worth sitting with: the test board itself now says it takes roughly double the vocabulary of “HSK 6-level” Chinese to reach the top of the scale, which is a fairly direct concession that HSK 6 alone was never C2-equivalent.

An honest HSK-to-CEFR mapping, with ranges instead of a single answer

Because HSK 7 through 9 only started real-world testing this year, any CEFR mapping for them is still provisional. Nobody, official or independent, has years of graded student outcomes to check them against yet. With that caveat stated plainly rather than buried in a footnote, here is a mapping that synthesizes the official framing, the independent European assessments, and the acknowledged uncertainty at the top of the new scale, given as ranges rather than a false single point:

Use the low end of each range if you want a conservative, defensible number for a CV or a university application. Use the high end if you are comparing against Hanban’s own promotional material. The honest answer for most learners sits closer to the low end, because the people who rejected the original 2010 mapping were the ones actually teaching the test’s graduates in a classroom, not the ones selling the test.

What this means for your own study plan

None of this changes how many hours it takes to get anywhere, it just changes what to call where you land. If your actual goal is a working professional level of Mandarin rather than a specific number on a chart, our fluency timeline calculator sidesteps the CEFR argument entirely and estimates your own timeline in hours, which is the unit that was never in dispute.

The other honest move is to stop trusting a vocabulary count as a proxy for your real level and go check it against actual text.

A vocabulary list can tell you how many words you have logged. Only a real passage can tell you whether they have actually turned into reading.