How Many Chinese Characters Do You Need to Know?
How many Chinese characters do I need to know? Sourced numbers from the HSK standard and frequency studies, and why words matter more than raw counts.
Chinese has tens of thousands of characters in big dictionaries, which sounds discouraging. The number you need in daily life is far smaller, and the numbers are surprisingly concrete once you look at how often characters actually appear. This article gathers figures from the HSK standard and from frequency studies, says where each comes from, and explains why the count alone can mislead.
What the HSK standard says
The current Chinese proficiency standard for international learners, Chinese Proficiency Grading Standards for International Chinese Language Education (GF 0025-2021), was issued in 2021. It has three stages and nine levels. Its word and character lists give these cumulative figures (as summarized on Wikipedia):
| Level | Words (cumulative) | Characters to recognize (cumulative) |
|---|---|---|
| 1 | 500 | 300 |
| 2 | 1,272 | 600 |
| 3 | 2,245 | 900 |
| 4 | 3,245 | 1,200 |
| 5 | 4,316 | 1,500 |
| 6 | 5,456 | 1,800 |
| 7 to 9 | 11,092 | 3,000 |
As a cross-check, counting the entries in an open-source dataset of the standard’s lists gives the same per-level numbers: 11,092 words and 3,000 characters in total. The standard also has a separate, smaller list of characters to write by hand (1,200 in total). The HSK exam syllabus released in November 2025 lowered the beginner and intermediate word counts, and learner sites now show tables such as this one with 300 words at level 1 and 5,400 at level 6. Use the numbers above as a rough scale and check the current official syllabus if you are preparing for a specific test.
Notice the pattern. The standard asks for about 1,800 characters through level 6 but over 5,000 words. Words outnumber characters about three to one, because characters get reused.
How much of real text do common characters cover?
Frequency studies count how often each character appears in large collections of text. The first few hundred characters do an enormous share of the work.
| Most frequent characters | Share of all characters in the text |
|---|---|
| 100 | 42.14% |
| 500 | 77.48% |
| 1,000 | 90.50% |
| 2,000 | 97.81% |
| 3,000 | 99.51% |
| 4,000 | 99.92% |
These figures come from Xing Hongbing (2007), as reported on Wikipedia: a corpus of 1,857,660 character occurrences containing 4,868 different characters. Other corpora give somewhat different numbers. Jun Da’s frequency list, in a version with data updated in 2004, shows about 64.6% cumulative coverage at rank 300 and 89.1% at rank 1,000, and a comparison table on YellowBridge gives 86.2% and 91.1% for the top 1,000 from two different counts.
The official Chinese lists tell the same story. The 1988 List of Frequently Used Characters in Modern Chinese (现代汉语常用字表, Xiàndài Hànyǔ Chángyòng Zì Biǎo) has 3,500 characters in two tiers. Per Wikipedia, citing the linguist Su Peicheng, testing found that the 2,500 most common cover 97.97% of text, the other 1,000 add 1.51%, and the total is 99.48%.
The practical lesson is the shape of the curve. Learning the first 1,000 characters gets you most of the way, and each additional 1,000 buys much less. But “covering 90% of characters” does not mean understanding 90% of a page. One unknown character in a ten-character sentence still blocks the sentence.
Characters are not words
Here is the biggest source of confusion. In modern Mandarin, the unit of meaning is often a word made of two characters. Knowing the characters is helpful, but it does not automatically give you the words.
- 火 (huǒ, fire) and 车 / 車 (chē, vehicle) combine into 火车 / 火車 (huǒchē, train).
- 电 / 電 (diàn, electricity) and 脑 / 腦 (nǎo, brain) combine into 电脑 / 電腦 (diànnǎo, computer).
Both examples are guessable once you know the parts, which is the real value of characters. See how Chinese compound words work.
English has the same split between parts and words. Brysbaert and colleagues (2016) estimated that an average 20-year-old American English speaker knows about 42,000 lemmas (base word forms), which come from roughly 11,100 word families. The count depends heavily on what you call a “word”. Chinese counts depend on definition too: 3,000 characters is a number about writing units, while 11,092 is a number about vocabulary entries. Comparing them directly mixes two different kinds of unit.
What to make of the numbers
- Around 1,000 characters account for about 90% of the characters in typical text, but you will still hit many unknown words.
- 2,000 to 3,000 characters is the range where reading real material becomes realistic, about 98 to 99.5% of characters in the studies above.
- Beyond 3,000 you are in diminishing returns territory; names, technical terms and classical Chinese call for more.
Also remember the difference between recognizing and producing. The HSK numbers above are for recognition. Writing every one of them from memory is a bigger task, and, as discussed in does handwriting help, you do not have to handwrite all of them. A common plan is to read many characters and write a smaller, well-chosen core.
Why a smaller, well-chosen set helps
It is tempting to treat “how many” as a race. A better question is which characters, in what order. A few hundred well-chosen characters can double as building blocks. 木 (mù, tree) appears inside 林 (lín, woods) and 休 (xiū, to rest), and many more characters reuse parts like 口 (kǒu, mouth) and 女 (nǚ, woman). Components reappear constantly, as covered in how characters are built and the comparison with English word roots. A list ordered so that you meet parts before the characters made from them lets each new character reuse what you already know.
I cannot give you a coverage percentage for any particular course list without counting it against a corpus, so treat claims like “this set covers X% of text” with suspicion unless the source is named.
Where Hanzibloc fits
Hanzibloc is an iPhone app with 70 units. You learn to write 700+ characters by hand, recognize 600+ more, and copy 30+ famous lines in copybooks. Each character is written stroke by stroke while the component you are writing lights up, and the aim is a solid base of common building blocks rather than a high count.
FAQ
How many Chinese characters do I need to read a newspaper?
Frequency studies of large text collections suggest that roughly 2,000 characters cover around 97 to 98 percent of the characters in typical modern texts, and 3,000 cover about 99 percent. You will still meet unfamiliar words, so dictionary use does not disappear.
How many characters does the new HSK require?
The 2021 Chinese proficiency standard lists 3,000 characters across its nine levels, with 1,800 of them in levels 1 to 6. The exam syllabus released in 2025 uses different word counts from the 2021 standard, so check the current official syllabus before an exam.
Is it better to learn characters or words?
Both, together. Most modern Chinese words are two characters, so you need words to actually communicate, but characters are the reusable parts that make new words easier to guess.