«How many characters do I need to read Chinese?» has a famous numerical answer, and the number is correct. The problem is that it misleads in two separate ways, and understanding both is worth more than memorising the figure.
Bill Edwards
·8 min read
Start with the good numbers, which exist and are official.
In 1988 the Chinese government published the 现代汉语常用字表, the list of frequently used characters in modern Chinese. There are 3,500, split into two bands for a very concrete reason:
| Band | Characters | Share of text covered |
|---|---|---|
| First level | 2,500 | 97.97% |
| Second level | +1,000 | +1.51% → 99.48% cumulative |
Look at the shape of that table, because it's the whole story: the first 2,500 give you nearly 98%, and the next 1,000 give you one and a half percent more. The effort multiplies and the payoff collapses.
Ninety-eight percent sounds like done. On a test it's an A. Translated into reading, it means one unknown character in every fifty — roughly one per line.
One stumble per line isn't reading. It's decoding with interruptions, and it's exhausting.
That's why people with 2,500 characters «known» open a newspaper and close it two paragraphs later. They don't lack characters: they need the missing one to stop showing up so often. And the jump for that isn't from 2,500 to 3,000 — it's somewhere else entirely, which is the second trap.
Modern Chinese is a language of two-character words, not single characters. And the meaning of the pair doesn't always follow from its parts:
| Word | Its parts | What it means |
|---|---|---|
| 电话 | 电 electric + 话 speech | telephone — you can work it out |
| 东西 | 东 east + 西 west | thing — you can't |
| 马上 | 马 horse + 上 on top | right away — you can't |
| 以为 | 以 with + 为 do | to think, mistakenly — you can't |
So you can know all 3,500 characters individually and still not get the sentence — the same way someone can know every letter in English and not know what deadline means. Counting characters measures the wrong unit. The unit of meaning is the word, and Chinese words run into the tens of thousands.
Characters stick through spaced repetition and stroke practice, not through lists. In Lemma they arrive animated and in your deck from lesson one.
The one that hurts is the alphabet. Somewhere in childhood you learned the letters, and from then on you could pronounce any written word even without knowing it. Decoding and understanding split into two steps, and the first one finished.
In Chinese that moment never arrives. There's no set of pieces that, once learned, lets you read anything aloud. If you don't know the character, you don't know how it sounds. Decoding and knowing are the same act, which is why «how many until I can read?» has no clean answer.
The one that helps is a habit you already have and don't think of as one: English taught you to read words by their roots. You know bio- is life, -ology is study, tele- is far — which is why biology and telephone don't feel arbitrary. Nobody taught them to you letter by letter. You took them apart.
Chinese is built exactly that way, across the entire language. 电话 is electric + speech. 火车 is fire + vehicle, train. 电脑 is electric + brain, computer. Where English keeps that logic for its learned vocabulary, Chinese uses it for everyday words.
And it goes one level deeper, inside the character. 好 (good) is 女 woman beside 子 child. 明 (bright) is 日 sun beside 月 moon. Characters aren't arbitrary drawings either: they're a couple of hundred components recombining. Once you see the pieces, you stop memorising shapes and start recognising combinations.
| Characters | What it actually lets you do |
|---|---|
| ~300 | Signs, numbers, prices, a menu. Getting around a city |
| ~1,000 | Close to 90% of text. Graded readers, simple messages |
| ~2,500 | The 98%. You can handle almost anything, with a dictionary and one stumble per line |
| ~3,500 | The literacy standard. Comfortable reading if you also have the words |
And the advice that actually matters: don't chase the number. Chase reading things. Frequent characters learn themselves if you read material that sits just above your level — and that material is the hard part to find, not the list.
Frequently asked
Around 3,000 to 3,500, which is the literacy standard in China. But characters alone aren't enough: you also need tens of thousands of words, because modern Chinese forms most of its vocabulary by combining two characters and the meaning of the pair doesn't always follow from the parts.
They're the first level of the 现代汉语常用字表, the official list of frequently used characters, and they cover 97.97% of ordinary text. The thousand in the second level add only 1.51% more, reaching 99.48%.
Around 300 will let you read signs, numbers, prices and a menu, which is what it takes to move around a city. Around a thousand gets you close to 90% of text and makes graded readers and simple messages workable.
Not enough, but it helps enormously. Characters are built from a couple of hundred components that recombine, so knowing them turns memorising shapes into recognising combinations. You still have to learn which combination means what.
You can learn to speak it, leaning on pinyin. But pinyin is an on-ramp, not a destination: real Chinese is written in characters, and without them you're locked out of everything written.
Animated strokes, spaced repetition and readings generated at your exact level — which is what closes the gap between beginner material and native text.
Keep reading
The consonants every guide tells you to fear are already in your mouth. The vowels you assumed you had are where you'll actually go wrong. Sorting that out in week one saves you a year of sounding off.
Pinyin isn't a pronunciation guide, it's a spelling system — and several letters don't mean what they do in English. The traps, the syllables that hide a sound, and the third tone you'll almost never say in full.
There are two different HSK exams using the same numbers, and most of what's written online describes the one that no longer applies. What each level means now, and why you probably don't need the exam at all.
The only serious number is 2,200 hours, and unlike most learners you're the exact person it was measured on. What it actually measures, why it still doesn't apply to you, and how long the goals you care about take.