You will constantly meet kanji you can see but cannot read — on a sign, a menu, a form. There are two questions that unlock them: how many strokes is this? and how do I search for it? A century ago the answer to the second depended entirely on the first. Today, handwriting input and OCR have largely replaced stroke-count lookup — but knowing how to count strokes still matters, and this page covers both the classic method and the modern shortcuts.
What counts as one stroke
A stroke (画（かく）, kaku; stroke count is 画数（かくすう）, kakusū) is a single, continuous movement of the pen from the moment it touches the paper to the moment it lifts. This is the definition English speakers most often get wrong, because a single stroke can change direction, hook, or bend without lifting — and every one of those is still one stroke.
- A bend (折れ, ore): the vertical-then-horizontal of コ-shaped strokes is one stroke, not two.
- A hook (はね, hane): the little flick at the end of the vertical in 手 or 子 is part of that stroke, not an extra one.
- A sweep (はらい, harai): a diagonal that tapers off is one stroke.
This is exactly why correct stroke order is the prerequisite for counting: if you know the standard order, you know which pen-lifts are real boundaries and which apparent "corners" are just a single stroke changing direction.
| Kanji | Reading | Strokes | Why that count |
|---|---|---|---|
| 川 | かわ (kawa) | 3 | Three separate verticals — three strokes, no bends |
| 口 | くち (kuchi) | 3 | Left vertical; top-and-right corner as one bent stroke; bottom horizontal |
| 力 | ちから (chikara) | 2 | The hooked bend is one stroke; the diagonal sweep is the second |
| 竹 | たけ (take) | 6 | Two mirror halves of three strokes each (3 + 3) |
| 曜 | よう (yō) | 18 | 日 (4) + 羽 (6) + 隹 (8) = 18 |
この川は三本の支流に分かれています。
kono kawa wa sanbon no shiryū ni wakarete imasu
This river splits into three tributaries.
竹はまっすぐ上に伸びます。
take wa massugu ue ni nobimasu
Bamboo grows straight up.
何曜日が一番忙しいですか。
nan-yōbi ga ichiban isogashii desu ka
Which day of the week is the busiest for you?
The classic method: radical + residual strokes
A traditional kanji dictionary (漢和辞典（かんわじてん）, kanwa jiten) is organized by radical (部首（ぶしゅ）, bushu). To find a character:
- Identify the radical — the classifying component, most often on the left, top, or enclosing the character.
- Look up the radical by its own stroke count in the radical index (部首索引（ぶしゅさくいん）, bushu sakuin). Radicals are ordered from 1 stroke (一) up.
- Find your character by the residual stroke count — the number of strokes left over once you set the radical aside.
So for 曜 you would go to the radical 日 (4 strokes), then look under residual count 14 (18 total − 4 for 日). Two other indexes exist for when the radical is unclear: the reading index (音訓索引（おんくんさくいん）, onkun sakuin) if you already know an on- or kun-reading, and the total-stroke index (総画索引（そうかくさくいん）, sōkaku sakuin), which lists every character purely by its full stroke count — your fallback when you know neither the radical nor a reading.
この漢字の部首は「さんずい」です。
kono kanji no bushu wa 'sanzui' desu
The radical of this kanji is 'sanzui' (the three-drops water radical).
読み方が分かれば、音訓索引で引けます。
yomikata ga wakareba, onkun sakuin de hikemasu
If you know a reading, you can look it up in the reading index.
The modern method: handwriting, OCR, and component search
Honestly, most learners today rarely open a paper dictionary. Three tools have made radical-and-residual lookup optional:
- Handwriting input. Switch your phone or computer IME to handwriting mode and draw the character. The recognizer uses your stroke sequence — one more reason stroke order matters — and offers candidates. See Typing and Input (IME).
- OCR / camera lookup. Point a dictionary app or translation camera at printed text and it reads the kanji for you. This is the fastest route for a character on a sign you cannot even begin to write.
- Multi-component (multi-radical) search. Apps like Jisho let you tap every component you recognize — say 日 and 隹 and 羽 — and they intersect those to find 曜. No stroke counting required.
知らない漢字はカメラで調べられます。
shiranai kanji wa kamera de shiraberaremasu
You can look up an unknown kanji with the camera.
画数が分からないときは手書き入力を使います。
kakusū ga wakaranai toki wa tegaki nyūryoku o tsukaimasu
When I don't know the stroke count, I use handwriting input.
So why still learn to count?
Three reasons stroke count keeps earning its place even in the age of OCR:
- Disambiguating look-alikes. Many near-identical characters differ by one stroke or by proportion. Knowing the exact count is often what separates them.
- Reading stroke-count-ordered material. The total-stroke index, kanji charts, and some name dictionaries are still sorted by 画数. You cannot navigate them blind.
- Names and culture. Japanese name-fortune-telling (姓名判断（せいめいはんだん）, seimei handan) is based entirely on the stroke counts of the characters in a name — parents really do choose kanji for a baby partly by stroke count. It is a live cultural practice, not a curiosity.
| Look-alikes | Readings | Strokes | The difference |
|---|---|---|---|
| 土 / 士 | つち (tsuchi) / し (shi) | 3 / 3 | Same count; 土 has the longer stroke on the bottom, 士 on top |
| 千 / 干 | せん (sen) / ほす (hosu) | 3 / 3 | 千 starts with a short sweep; 干 with a flat top bar |
| 未 / 末 | み (mi) / まつ (matsu) | 5 / 5 | 未 has the shorter bar on top; 末 the longer bar on top |
| 大 / 太 / 犬 | だい (dai) / ふと (futo) / いぬ (inu) | 3 / 4 / 4 | 太 adds a dot below, 犬 a dot on the upper right |
「土」と「士」は画数が同じですが、形が少し違います。
'tsuchi' to 'shi' wa kakusū ga onaji desu ga, katachi ga sukoshi chigaimasu
'土' and '士' have the same stroke count, but their shapes differ slightly.
子供の名前の画数を気にする親も多いです。
kodomo no namae no kakusū o ki ni suru oya mo ōi desu
Many parents pay attention to the stroke count of their child's name.
Common mistakes
❌ 口 を4画と数える
kuchi o yon-kaku to kazoeru
Incorrect — counting 口 as four strokes.
✅ 口 は3画（縦・折れ・横）
kuchi wa san-kaku (tate, ore, yoko)
Correct: 口 is three strokes — the top-right corner is one bent stroke, not two.
❌ 手 のはねを別の一画と数える
te no hane o betsu no ikkaku to kazoeru
Incorrect — counting the hook of 手 as a separate stroke.
✅ はねは同じ一画の一部
hane wa onaji ikkaku no ichibu
Correct: a hook is part of the same single stroke.
❌ 力 を3画と数える
chikara o san-kaku to kazoeru
Incorrect — counting 力 as three strokes.
✅ 力 は2画（折れ・はらい）
chikara wa ni-kaku (ore, harai)
Correct: 力 is two strokes — one bent-and-hooked stroke plus one sweep.
❌ 部首を無視して総画だけで引こうとする
bushu o mushi shite sōkaku dake de hikō to suru
Incorrect — ignoring the radical and trying to look everything up by total strokes.
✅ まず部首、次に残りの画数で引く
mazu bushu, tsugi ni nokori no kakusū de hiku
Correct: search by radical first, then by the residual stroke count.
Every one of these traces back to the same root cause: not knowing where one stroke stops. Fix that by anchoring your counting to stroke order and to components, and both the number and the lookup fall into place.
Key takeaways
- One stroke = one continuous pen movement; bends, hooks, and sweeps do not add to the count.
- Classic lookup is radical → residual stroke count; total-stroke and reading indexes are the fallbacks.
- Handwriting input, OCR, and multi-component search have made radical lookup optional for daily use — but stroke count still disambiguates look-alikes and underpins name culture.
Now practice Japanese
Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.
Start learning Japanese→Related Topics
- Stroke Order (筆順) BasicsN5 — The handful of ordering rules that make every kanji legible, correctly proportioned, and recognizable to handwriting input — and why they still matter in a typed world.
- Radicals (部首): The Building Blocks of KanjiN4 — How kanji decompose into recurring components — semantic radicals that hint at meaning and phonetic components that hint at the reading — and why two-thirds of kanji are phono-semantic compounds you can partly predict.
- Typing Japanese: The IMEN4 — How you type three scripts on one keyboard — romaji becomes kana, the space bar converts kana to kanji candidates you choose — and why the IME makes you spell は as 'ha', を as 'wo', and づ as 'du'.