What Order Should You Learn Chinese Characters In?
Frequency lists, HSK bands, and component dependency all claim to be the right sequence. Only one of them survives the first month, and it isn't the one most apps use.
Most people start learning Chinese characters in whatever order their app or textbook hands them, and never ask where that order came from. It matters more than almost any other decision you make, because a bad sequence forces you to memorize shapes you have no way to reason about.
Short answer: learn characters in frequency order within a difficulty band, and inside that band put components before the characters built from them. Pure frequency order is too chaotic; pure component order delays words you need in week one.
The four orderings below are the ones that actually get used. Each is defensible on its own terms. The differences show up around character 200, which is roughly where most people quit.
The Four Orders People Actually Use
| Ordering | Sequenced by | Strength | Where it breaks |
|---|---|---|---|
| Frequency | How often a character appears in a corpus | Highest early reading payoff per character | Ignores visual complexity entirely; 的 and 我 arrive beside 意 and 题 |
| Textbook / HSK | An official syllabus or course tree | Matches an exam, and matches a class | Optimized for spoken topics, not for how characters are built |
| Component dependency | Simple shapes before the compounds using them | Every new character decomposes into known parts | Some components are rare as standalone characters, so they'd arrive absurdly late |
| Encounter order | Whatever you meet while reading or watching | Everything is contextual and memorable | No coverage guarantee; you meet 咖啡 before 是 |
The fourth is not really a plan, but it is what happens by default if you learn from content rather than a list. It works for some people and leaves enormous gaps for most.
Why Pure Frequency Order Breaks Down
Frequency order is seductive because the maths is unarguable. The 100 most common characters cover a large share of running text, so front-loading them buys reading ability fast. Our own frequency list is built on exactly that logic.
- Complexity is uncorrelated with frequency. 意 (yì, meaning) is more common than 山 (shān, mountain), and has 13 strokes to 山's three.
- Grammatical particles dominate the top of the list. 的, 了, 是, and 在 are the most frequent characters in Chinese and among the least visually informative, so your first week is spent on shapes that teach you nothing about the system.
- Components arrive after their compounds. In a strict frequency list, 明 (míng, bright) shows up well before 月 (yuè, moon), so you memorize a shape whose left half you could have read.
Frequency order optimizes for coverage per character. It does not optimize for effort per character, and the second one is what determines whether you're still studying in March.
Why Component Order Is Better -- And Where It Fails
Chinese characters are not arbitrary. Across the 3,282 characters in HSK 1 through 9, 60.1% are pictophonetic -- one part carries a meaning hint, the other a sound hint -- and another 25.3% are ideographic compounds, where two meaning parts combine. Only 4.2% are pictographs you simply have to recognize.
That distribution is the whole argument for component-first sequencing. If you know 日 (rì, sun) and 月 (yuè, moon), then 明 (míng, bright) is not a new shape to memorize -- it is two shapes you already read, in the most common arrangement there is.
| Character | Pinyin | Parts you already know | What the parts tell you |
|---|---|---|---|
| 明 | míng | 日 sun + 月 moon | Two light sources, side by side: bright |
| 好 | hǎo | 女 woman + 子 child | An ideographic compound meaning good |
| 河 | hé | 氵 water + 可 kě | Water on the left, sound on the right: a river, read hé |
| 清 | qīng | 氵 water + 青 qīng | Water plus a sound part: clear, read qīng |
| 林 | lín | 木 tree + 木 tree | Two trees: a wood |
The failure mode is real, though. Strict dependency ordering says you cannot meet 的 (de) until you have met 勺 (sháo, ladle) -- and 勺 is rare, so in a frequency-sorted curriculum it lands hundreds of characters later. Nobody should wait that long to learn the single most common character in the language.
The Compromise That Actually Works
- Bound the sequence by difficulty band first. HSK levels are a coarse but workable proxy: they keep 你好 in week one and 概念 in year two.
- Sort by component dependency inside each band. Where both a component and its compound sit in the same band, teach the component first.
- Break remaining ties by frequency. Two equally decomposable characters at the same level: the more common one goes first.
- Allow deliberate exceptions for the top of the frequency list. 的, 是, 在, and 了 come early regardless of what their parts are, because they are unavoidable.
This is a hierarchy, not a formula. When the difficulty band and the dependency graph disagree, the band wins -- and it disagrees about a quarter of the time.
That figure is worth being precise about. Measured across the 1,940 characters in HSK 1 through 6, roughly 73% of characters that have a component also appearing in the sequence meet that component at or before their own unit. The remaining 27% are cases where the component is itself a rare character, and forcing it early would be worse than the inconsistency.
A Worked Example: The First Six Units
Here is what the compromise produces in practice, at five characters per unit. Notice 日 and 月 in the first two units, which is what makes 明 readable by unit 17.
| Unit | Characters | What it sets up |
|---|---|---|
| 1 | 一 口 日 白 的 | 日 and 白 arrive together, so 的's left half is a known shape |
| 2 | 是 不 了 二 月 | 月 lands early, pairing with unit 1's 日 |
| 3 | 有 在 文 这 手 | 手 is the standalone form of the 扌 radical, which heads 190 characters |
| 4 | 我 人 们 也 他 | 人 immediately before 们 and 他, both of which use the 亻 form |
| 5 | 个 来 到 上 说 | 说 introduces 讠, the speech radical, which heads 82 characters |
| 6 | 女 要 会 大 时 | 女 arrives before 好 in unit 12 |
Three of those six units front-load a radical that heads a large family. That is not a coincidence -- it is the highest-leverage move available, because one component learned in unit 3 pays off across dozens of later characters.
What This Means For Your Study Plan
Learn radicals as characters
Decompose before you drill
Let review handle retention
If you are building your own list, the practical shortcut is to take an HSK level, group its characters by radical, and study each radical's group together. You lose the strict frequency ordering and gain far more than you lose.
HanziFeed does this as its default sequence: 391 units of about five characters each, HSK 1 through 6, ordered by component dependency inside each band with frequency breaking ties. Every unit shows the parts before it asks you anything.
Frequently Asked Questions
Should I learn all 205 radicals before starting characters?
Is HSK order good enough on its own?
What about learning characters in traditional order, from classical texts?
Does order matter if I'm only learning to read, not write?
How many characters should be in one study session?
Study characters in an order that builds on itself
HanziFeed sequences 1,940 characters into 391 short units, component before compound, with a checkpoint every fifth unit you can skip if you already know the material.