Chinese vocabulary and HSK FAQ

This is the single most confusing fact in the field, and the confusion is legitimate rather than anyone's mistake. The nine-level HSK 3.0 standard was announced by Chinese Testing International (formerly Hanban) in 2021 and is fully official, and some test centers now administer it. But the rollout has been staggered across several years, and many centers still run the old six-level HSK 2.0 exam. A large share of apps, textbooks, and published word lists also have not been updated. The result is that both structures are genuinely current in the wild, which is why one page tells you there are six levels and the next says nine, and both are describing a real exam. The practical answer to which one applies to you is: whichever your specific test center administers, so ask them directly rather than assuming from a search result.

Split the decision by why you are studying. If you are working toward a certificate, the version is not a preference, it is dictated by your test center, so confirm directly whether the sitting you will take is the six-level or nine-level exam and study to that structure; preparing against the wrong one is the one avoidable way to waste the fee. If you are learning for yourself with no exam booked, the choice barely matters, because the two standards overlap heavily at the beginner end and the most frequent 1,000 words serve you well under either. The pragmatic path for a self-learner is to ignore the version question at first, learn high-frequency vocabulary in frequency order, and only pin down a standard later if and when you decide to certify. Either way, note which version any level number you see came from, since the same number means different word counts on each ladder.

It depends on where on the ladder you look. HSK 3.0 kept more, smaller-looking steps (nine levels instead of six) but raised the vocabulary counts at every stage, most dramatically at the top. The beginner band spans three levels of its own and finishes at 2,209 words in our published 3.0 list, more than the old exam's 600 at HSK 3, so it is not actually lighter, just spread across more steps. At the top it is dramatically harder, reaching 10,969 words versus 5,000 for old HSK 6, and it folds in a speaking test that becomes mandatory from Level 3, where the old HSKK spoken exam was separate and usually skipped. The steepest increases are in the intermediate and advanced bands. (Some summaries quote rounded early syllabus targets, a 1,000-word beginner band or 11,000 at the top; those run below the published word list.)

This is one of the bigger practical changes between the two standards, and it catches people out. Under the old six-level HSK 2.0, the spoken test, the HSKK (Hanyu Shuiping Kouyu Kaoshi), was administered as a completely separate, optional exam. You could sit HSK 2.0 and earn a certificate on reading and listening alone, never speaking a word, which is why so many older certificates say nothing about spoken ability. The nine-level HSK 3.0 standard folds speaking into the single graded framework and makes it a required component from the intermediate levels, so from Level 3 upward you can no longer certify without demonstrating speaking. This matters for how you study: if you are preparing on the new standard, you cannot treat speaking as a bolt-on to add later, and listening-heavy, silent study will leave you short at the exam. Our guide on the HSKK speaking test covers the format, and the HSK 2.0 versus 3.0 guide lays out every structural change between the two.

The reason you see different answers is that there are two live standards, and even the 3.0 figure is quoted two ways. The current nine-level HSK 3.0 standard, announced in 2021, sets Level 1 at 506 words in our published 3.0 list. The older six-level HSK 2.0, still administered at many test centers, put HSK 1 at 150 words, so the figure more than tripled between versions. A rounded early target of 300 words for 3.0 Level 1 also circulates, but the published word list runs larger. Note that these are word counts, not character counts; HSK 3.0 Level 1 corresponds to roughly 246 recognition characters, and characters and words are not interchangeable because many Chinese words are built from two or more characters. If you are studying toward a certificate, confirm which standard your test center uses before you trust a Level 1 word list, because the same level number means different things on each ladder.

Level 4 is the entry point of the intermediate band on both standards, but the count moved. Under HSK 3.0 it certifies 3,181 cumulative words in our published 3.0 list, adding 972 new words on top of the 2,209-word beginner band. Under the old HSK 2.0, HSK 4 required 1,200 cumulative words, so the newer standard asks for far more. This is why so many self-studiers describe hitting a wall around here: the intermediate band adds more than 3,000 words on top of the beginner band and the vocabulary turns abstract, less concrete than the everyday nouns and verbs that came before. The fix is a lower, steadier pace through this band, not a sprint. (Some summaries quote rounded early targets, a 2,000-word Level 4 and a 1,000-word beginner band; those run below the published word list.)

Fluency is not a level, so no HSK number certifies it outright, but the framework does mark where working proficiency begins. Through the beginner band (up to Level 3, 2,209 words) you can handle predictable everyday situations, which is useful but not what most people mean by fluent. Real working proficiency starts in the intermediate band: by Level 5 to 6 in HSK 3.0 terms, 4,240 to 5,363 words, you can read general material and discuss abstract topics, and this is the range universities and employers usually treat as a working benchmark. Near-native reading, including literary and specialized language, lives in the advanced band (Levels 7 to 9, 10,969 words). Keep in mind that vocabulary size is only one dimension; genuine fluency also needs listening, speaking, and time using the language, none of which a word count measures.

Effective vocabulary learning comes down to three levers, in order: what you learn, how you review it, and how you encode it. Learn the most frequent words first, because a small set of common words does most of the work in real text, so frequency order buys the fastest usable payoff per word. Review with spaced repetition so each word resurfaces just before you would forget it and you spend time only on words at risk, not ones you already know. Encode every word inside an example sentence rather than as a bare character-to-English gloss, because context fixes usage, tone, and the measure word and makes the word show up reliably in real reading. Set a daily new-word count you can sustain, protect your due reviews before adding new cards, and expect the intermediate band to take the most time.

The trap is picking a new-word rate based on a good day rather than an average one. Every new word you add returns as future reviews, so the real daily load is your new words plus everything already due, and that review pile grows for weeks before it stabilizes. Most self-studiers settle comfortably around 5 to 15 new words a day; 10 is a sensible default that clears the 2,209-word beginner band in roughly seven to eight months. The right choice is the number you can repeat on a tired day, not the maximum you can manage on a fresh one, because consistency compounds and cramming leaks straight back out. If your reviews start feeling unmanageable, cut new words before you cut reviews: protecting the words you already have is worth more than adding new ones.

Level 3 sits at the top of the HSK 3.0 beginner band, which certifies 2,209 cumulative words in our published 3.0 list, so the vocabulary math is straightforward once you fix a daily rate. At 10 new words a day, holding your reviews steady, that is roughly seven to eight months; at a gentler 5 a day it stretches past a year. Two honest caveats keep this from being a promise. First, the number assumes you actually keep the reviews current rather than piling up a backlog, which is where most timelines quietly break. Second, HSK 3 is not only vocabulary: grammar, listening, and (from Level 3 on the new standard) speaking all take practice the word count does not capture. Treat the vocabulary timeline as a floor for a focused learner, then add time for the skills around the words. (Some summaries still cite a 1,000-word beginner band from an early syllabus target; the published word list is larger.)

The scheduling engine is the same idea in both, so this is a question of setup cost, not effectiveness. Anki wins on price (free on most platforms) and control: you can build exactly the deck you want, add your own audio, and tune the algorithm. The cost is time spent sourcing or importing HSK decks, cleaning them, and maintaining them. A purpose-built Chinese app trades that flexibility for a running start, shipping HSK-tagged word lists, example sentences, and audio out of the box so you spend your minutes reviewing rather than configuring. The honest test is which friction you will actually tolerate every day: if fiddling with cards energizes you, use Anki; if it is the thing that would make you quit, use an app that is ready on install.

You cannot read your level off a word count, because recognizing a word sitting on a list next to its English gloss is a much easier task than recalling it cold inside a real sentence, and that gap is why self-estimates run high. The reliable method is a short recognition check that samples words from several bands rather than testing one level in isolation. What you are looking for is the accuracy crossover: the band where you fall from comfortable (around 90 percent) to shaky (below 75 percent). That transition point, not a round vocabulary number, is your working level. One extra step matters: translate the result onto the ladder your exam actually uses, because the same vocabulary maps to different level numbers on HSK 3.0 (nine levels) versus HSK 2.0 (six levels). A few minutes of honest testing beats guessing and points you straight at the words worth drilling next.

HSK is really two things: a paid certificate and a free framework, and only the certificate is ever mandatory. You need the exam if something external asks for it: admission to a degree taught in Chinese, a government or university scholarship, a work visa, or a China-facing job that lists an HSK level as proof of proficiency. A booked exam can also work purely as a deadline that keeps your study honest, which is a legitimate reason to pay for one. If none of those apply and you are learning for travel, family, or interest, you can skip the test entirely without skipping HSK: the level structure and frequency-ordered word lists answer the hardest self-study question, what to learn next, and cost nothing to follow for years.

The best first words are the ones you cannot avoid using. That means the structural core of the language: pronouns (I, you, he, she), numbers, the highest-frequency verbs (to be, to have, to want, to go, to like), everyday nouns (person, water, food, home, day), and question words (what, who, where, how many). This set lines up closely with HSK 3.0 Level 1, which is about 500 words, and it is worth learning first for a concrete reason: a few hundred common words account for a large share of everyday text, so each one you learn early shows up constantly and pays off immediately. Learn them in context inside short sentences rather than as bare glosses, because words like measure words and particles only make sense in use, and starting from frequency means you are always working on the word most likely to appear next.

The confusion here almost always comes from mixing up characters and words. Under the current HSK 3.0 standard, each of Levels 1 to 6 adds close to 300 new characters, so Level 6 reaches 1,799 distinct characters cumulatively, and the advanced 7 to 9 band pushes that to about 2,970. HSK itself is scored on words, not characters, and the two totals are not interchangeable: one word can reuse characters you already know, so the character count always climbs more slowly than the word count. Our characters-per-level tables show the full breakdown for both schemes side by side, computed from the shipped HSK word lists with radical and stroke data from the Unicode Han Database. If you are looking at an old six-level HSK 2.0 chart the character figures are completely different, front-loading fewer characters early and adding far more per level later, so always check which standard a count refers to before you trust it.

A common beginner trap is to try to memorise all 214 radicals up front as a wall of abstract shapes, which is slow and demotivating because nothing is anchored to a word you can use. The better approach is to learn radicals in the flow of learning characters: when a new character shares a part you have seen before, name that part and move on. The payoff is lopsided in your favour because a small number of radicals do most of the work. On our radicals explorer you can see the 214 Kangxi radicals ranked by how many HSK characters use each: the single hand radical indexes 185 HSK characters, and the top eight together account for a large share of everything you will meet. Understanding that most characters split into a meaning part and a sound part turns each new character from a fresh shape into a small combination of pieces you already recognise. So learn the high-frequency radicals early, but learn them attached to characters, not in isolation.

People use the two words loosely, but they are not the same thing, and the distinction clears up a lot of confusion. A component is any recurring sub-shape a character is built from: the character 好 is made of two components, 女 (woman) and 子 (child). A Kangxi radical is narrower: it is the single component the dictionary uses to file that character, the one that decides where it lives in the index. So every radical is a component, but most characters contain several components while only one of them is the indexing radical. The distinction matters in practice because a component that is a full radical in one character can appear as a plain sound or meaning part in another, and some radicals also take a reduced variant form inside compounds. Our radicals explorer lists all 214 Kangxi radicals and the HSK characters filed under each, and the characters reference shows how they combine across every HSK level.

This is one of the most argued-over questions in Chinese-learning communities, and the honest answer is that it depends on your goal and your test format. For everyday digital life you type Chinese with pinyin and select the right character, which leans entirely on recognition rather than production, so recognition is the higher-leverage skill for most learners. The HSK reflects this split: the internet-based version has you type, while the paper version still asks you to handwrite characters in the writing section at the upper levels, so if you plan to sit the paper test you cannot skip handwriting entirely. Beyond the exam, there is a real memory argument for writing: physically producing a character forces you to attend to its components and stroke order in a way passive recognition does not, which can make characters stick. A practical middle path many learners take is to prioritise reading and typing for coverage, while hand-writing the highest-frequency characters to lock them in. Our characters reference and radicals explorer help you see the shared parts that make writing less of a memory load.

The HSK is run by the mainland Chinese testing authorities, and it is administered entirely in simplified characters, the writing set standardised in mainland China. Traditional characters, still used in Taiwan, Hong Kong, and Macau, do not appear on the exam, so for HSK preparation you only need simplified. This is why every character and word list on this site is in simplified form. That said, the two sets overlap heavily: a large share of characters are identical in both, and many differences are systematic, so learners who already know simplified can pick up traditional later with far less effort than starting over. If your longer-term goals include reading pre-1950s texts, classical literature, or content from Taiwan and Hong Kong, treat traditional as a worthwhile second layer once your simplified reading is solid, not a competing first choice. For the exam itself, and for most communication with the mainland, simplified is the clear starting point. Our HSK vocabulary tables and characters reference are all simplified.

Coverage figures are genuinely useful, but they are easy to over-read, so it helps to know exactly what they measure. Token coverage is the share of running words in real speech or text that your vocabulary would recognise, counting every occurrence. Because a small set of very common words repeats constantly, coverage climbs fast early and then flattens: HSK 3.0 Level 1 already recognises about half of spoken words, and by Level 5 you reach roughly 72 percent, rising to about 74 percent at Level 6. But recognising a word is not the same as understanding a sentence, which is the distinction between comprehension and coverage. The unknown words are often the exact ones carrying the point, and coverage says nothing about grammar, idiom, cultural reference, or the speed of real-time listening. So treat coverage as a floor that tells you roughly how much you will meet, not a ceiling that tells you how much you will understand. Our coverage page shows the full per-level curve for both HSK schemes, computed against a published spoken-register frequency list.

There is no single magic number, because watching television leans on listening speed and idiom as much as on vocabulary size, but the coverage curve gives a useful guide. This site measures token coverage against a film and television frequency list, which is close to what you actually hear on screen. By HSK 3.0 Level 6, roughly 5,000 to 5,500 words, you recognise about 74 percent of spoken words, which is enough to follow the gist of many everyday dramas and talk shows, especially with visual context. But 74 percent recognition still means a meaningful share of words fly past unknown, so comfortable, subtitle-free viewing of a wide range of content really belongs to the advanced 7 to 9 band, which adds thousands more words. Two honest caveats: coverage is not comprehension, and genre matters enormously, since a slice-of-life sitcom is far more forgiving than news or a period drama. See the coverage page for the full per-level breakdown, and the placement test to find where you currently stand.

A widely cited rule of thumb, echoed across Chinese-learning forums, is that knowing the 2,500 to 3,000 most frequent characters lets you read the great majority of a modern newspaper, with the top 3,000 or so characters covering the overwhelming share of characters in everyday printed text. It is worth separating two counts here, because people conflate them: a character coverage figure counts distinct characters, not words, and the two totals never line up. On the HSK 3.0 ladder you reach 1,799 distinct characters by Level 6 and about 2,970 across the advanced 7 to 9 band, so that top band lands right in the functional-news-reading range. Two caveats keep this realistic. First, coverage is not comprehension: recognising the characters does not guarantee you parse the sentence, especially in dense political or financial writing. Second, proper names, technical terms, and idioms will keep tripping you up well past 3,000 characters. Our characters reference shows the distinct-character count for every HSK level, and the coverage page shows how quickly the common core does most of the work.