RubricMark · Study Lab

How many Chinese characters are there

The 2013 official table has 8,105 characters: 3,500 common, 3,000 next, 1,605 specialised. Kangxi and Unicode are different counts.

There is no single number of Chinese characters. There is a list for school and ordinary reading, and there are much larger counts from dictionaries and from computers. Mixing them is how “how many Chinese characters” turns into a project nobody finishes.

The list to use is the Table of General Standard Chinese Characters, published in 2013 by the Ministry of Education and the State Language Commission. It has 8,105 characters in three levels.

3,500, then 6,500, then 8,105

Level 1 has 3,500 characters. That is the common set, the one meant for basic education and general literacy. It is not “grade 1”, and it is not an alphabet. Level 2 has 3,000. Levels 1 and 2 together are the 6,500 characters described as enough for ordinary publishing and general use. Level 3 has 1,605: names, place names, technical terms, and characters from classical texts in the school curriculum that did not make the first two levels. 3,500 + 3,000 + 1,605 = 8,105.

[[chart:ld-hanzi-scale]]

Stay on the first view if you are deciding what to study. Switch to the second view only to see why other numbers keep appearing. The axis is logarithmic there, because 3,500 and 101,996 cannot share a linear chart without the study list disappearing. The log scale is a warning, not a suggestion that those four bars are the same kind of thing.

The totals that are not a study plan

The Kangxi Dictionary (1716) states 47,035 headwords. That is a dictionary’s own count, three centuries old, including characters you will not meet in a newspaper. Unicode 17.0, released on 9 September 2025, encodes 101,996 CJK Unified Ideographs — characters from Chinese, Japanese, Korean, and Vietnamese sources, unified when the consortium treats them as the same character. The Unicode statistics page shows a Han subtotal of 102,998 because it also counts 1,014 compatibility ideographs. Neither figure is a reading list.

HSK word totals get pulled into the same sentence. The exam’s older six-level description, still the one on official explainers, is about 150, 300, 600, 1,200, 2,500, and over 5,000 words. 谢谢 is one vocabulary item and two characters. Do not draw a bar of “5,000 characters” from that line. When you want the practice test for your exam, it is the HSK sit — after the characters you actually need are retrievable, not before.

Where a card should stop

A deck that aims at Unicode is a code chart. A deck that aims at level 1 is a school list. Clear the 3,500 before you treat level 2 as urgent, and treat level 3 as the set you add when a name or a classical line actually shows up. The same discipline on the Japanese side is how many kanji, and the reason those lists are not interchangeable is hanzi, kanji, and hanja.

Why this method works

A list you can name (“eight thousand”) is not a list you can read. The 2013 table is useful because it stops. Level 1 is a closed set of 3,500. You can tell, on a given day, whether a new character is inside that set or a specialist extra. Open sets — Kangxi, Unicode — cannot be finished, so they never become a session.

Run this session

Twenty-five minutes. Do not download a “all characters” deck.

  1. 0–5 min. Write the three levels from memory: 3,500, 3,000, 1,605. Check the first chart view. If you wrote 47,035 or 101,996 as the study target, that is the second view, and it is the wrong job.
  2. 5–20 min. Ten characters that appear in ordinary sentences: 的, 了, 是, 不, 我, 人, 有, 他, 这, 中. These are examples, not an official top-ten ranking. Front: character. Back: pinyin with tone, and one word. Cover. Produce. Any miss stays.
  3. 20–25 min. One HSK trap, on paper: write “谢谢 is ___ word(s) and ___ character(s).” The answer is one word, two characters. If your deck counts words and characters as the same bar, split it today.

Common traps

  • Using the Unicode total as a progress bar. Fix: progress is level-1 characters you can produce, not code points you have seen.
  • Calling level 1 “grade 1”. Fix: it is the common set. Japan’s grade 1 is 80 kanji, a different list.
  • Quoting HSK 6 as 5,000 characters. Fix: the exam line is words, and level 6 is published as over 5,000 words.

Produce the character from the word, or the word from the character, in Study Lab. A total you have only read is not a character you can write.