Stroke count is the number of strokes it takes to write a character, and it is used for one thing above all: finding a character in a dictionary. It is a poor measure of difficulty, a poor guide to learning order, and a genuinely useful piece of information once you know what it is for.
What counts as one stroke
A stroke is a single continuous movement of the pen from the moment it touches the paper until it lifts. That includes movements that change direction: the 乙-shaped stroke that turns twice is still one stroke, because the pen never leaves the page.
This is why stroke counts surprise learners. 口 is three strokes, not four, because the top and right side are written as one movement that turns a corner. 田 is five for the same reason.
The counting convention is standardised, which is what makes it usable as an index. Dictionaries, character databases and the Unicode Han database all use the same counts, so a character has one canonical number rather than one per source.
| Character | Strokes | The catch |
|---|---|---|
| 口 | 3 | Top and right side are one stroke |
| 田 | 5 | Same corner-turning rule |
| 木 | 4 | Straightforward |
| 女 | 3 | The first stroke turns twice |
| 車 | 7 | Easy to overcount |
What stroke count is for
Dictionary lookup, first and foremost. The traditional method finds a character by its indexing radical, then by how many strokes remain outside that radical, which is why both numbers matter.
Sorting, second. Character lists, indexes and reference tables order by stroke count because it is objective and stable, which alphabetical ordering cannot be for a non-alphabetic script. The jōyō kanji table and most reference material use it for exactly this reason.
Font and typesetting work, third, and irrelevant to most learners.
What it is not for: deciding what to learn. Stroke count correlates weakly with frequency, usefulness or difficulty, and ordering study by it front-loads simple rare characters while postponing complex common ones.
Why high stroke counts are not hard
This surprises people, and it holds up in practice. A character with eighteen strokes built from four components you already know is easier to write than a nine-stroke character with an unfamiliar structure.
The reason is that you do not write strokes, you write components. 議 has twenty strokes and two parts, 言 and 義, both of which a learner meets early. Once those are automatic, the character is two movements of familiar territory rather than twenty separate decisions.
The practical implication for study order is the opposite of the intuitive one: learn frequent components early, and complex characters stop being intimidating on their own.
Difficulty comes from unfamiliar structure, unusual proportion, and similarity to another character you already know, in roughly that order. Stroke count is barely in the list.
Where stroke count does affect writing
Two places, both about handwriting rather than learning.
Proportion under pressure. A high-stroke character has to fit the same square as a low-stroke one, so its internal spacing is tighter, and cramping is the most common visible error. This is a real skill and it is trained by writing large first.
Speed and legibility. More strokes take longer, and at speed a complex character degrades faster than a simple one, which is where automatic stroke order earns its keep.
Neither is a reason to avoid complex characters. Both are reasons to practise them at size before writing them quickly.
| Common belief | Reality |
|---|---|
| High stroke count means hard | Unfamiliar structure means hard |
| Learn low-stroke characters first | Learn frequent components first |
| Stroke count predicts study time | Component familiarity predicts it |
| Complex characters are rare | Many high-count characters are everyday words |
The number that would be more useful
If stroke count is a weak proxy for difficulty, it is worth asking what a good one would look like.
Component count comes closer: a character made of two familiar parts is easier than one made of five unfamiliar ones, regardless of strokes.
Neighbour density is closer still. A character with several visually similar relatives is harder to produce reliably, because the difficulty is discrimination rather than complexity, and this is where most persistent errors live.
Personal familiarity beats both, and it is the one nobody can publish, because it depends on what you have already learned.
That is why difficulty ratings in learning materials are approximate at best, and why your own record of what keeps failing is more useful than any published ordering. The list of characters you get wrong is a difficulty rating calibrated exactly to you.
Using stroke count practically
Count when you need to look something up in a paper dictionary or a stroke-ordered index. That is its job and it does it well.
Use it as a sanity check on your own writing. If your count for a character differs from the standard, you are probably combining or splitting a stroke, which usually means your stroke order has drifted.
Ignore it when choosing what to study. Frequency, your textbook and the vocabulary you actually meet are all better guides.
And do not use it to estimate effort. A learner who has met 言 twenty times will write 議 faster than an unfamiliar seven-stroke character, which is the whole argument for structural learning.
Stroke count and difficulty, tested
If you doubt that stroke count is a weak difficulty signal, the test takes two minutes.
Pick a high-count character you use often, and a low-count character you never see. Write both from memory. Almost everyone finds the familiar complex one easier, because familiarity and structure beat simplicity.
Then try a pair that differs only in one component, such as characters sharing a phonetic part. Those are harder than either of the first two despite unremarkable stroke counts, because the difficulty is discrimination rather than complexity.
That result is worth internalising, because it redirects effort. Time spent avoiding complex characters buys nothing; time spent on the pairs you confuse buys a great deal.
Difficulty in kanji is mostly about similarity and unfamiliarity. Stroke count measures neither, which is why an index built on it is excellent for lookup and useless for planning.
The lookup skill, briefly
Radical-plus-stroke-count lookup is worth knowing even in an era of digital tools, because dictionaries and reference material are organised around it.
The method: identify the indexing radical, find that section, then locate the character by the number of strokes outside the radical. It requires you to count correctly, which is why the corner-turning rule matters.
Digital alternatives are usually better. Multi-radical search lets you pick several components you can see, which forgives not knowing which one is the official index. Handwriting input works when you can approximate the shape, and it doubles as a test of whether you can produce it.
The one situation where counting still wins is a printed dictionary with no device to hand, which happens more often than it sounds if you study from paper.
{{appName}} does not ask you to count strokes; it checks the order of the ones you write, prompt first with the character hidden, and returns the ones you miss sooner. It is free in early access. For the order itself, remembering stroke order and learning kanji strokes go deeper.
Where the count comes from
The standardisation is worth a paragraph, because it explains why sources agree.
Stroke counts follow the shapes of characters as they are conventionally written, which is why they are tied to stroke order rather than to how a character looks in a particular font. A stroke is defined by the movement, not by the visual segment.
That is also why a printed character can appear to have more visible marks than its official count: a turning stroke drawn with a sharp corner looks like two lines and is one movement.
Because the convention is shared, every major dictionary, character database and educational resource reports the same number for a given character, which is what makes stroke count usable as an index at all. A count that varied by source would be worthless for lookup.
For a learner the practical takeaway is small: if your count disagrees with a reference, the reference is right and your stroke order is worth checking.
What to use instead for planning
If stroke count is the wrong planning tool, something has to replace it, and three things do.
Frequency, which decides how often you will meet a character and therefore how much free reinforcement it gets. This is the single best ordering principle for reading.
Component familiarity, which decides how expensive a character is to learn. A character made of known parts is cheap regardless of its stroke count.
Personal relevance, which decides whether you will actually retain it. The characters in your own name, address, work vocabulary and interests are reinforced by life rather than by study.
Most published learning sequences encode the first two implicitly, which is why following a textbook or school order beats inventing an order from a character property. The Unicode Han database is where to look if you want the raw structural data behind a character, including its components and its canonical stroke count.
When counting still matters day to day
Three small situations where a learner genuinely uses the number.
Looking something up in a printed dictionary or an index, which is the original purpose and still the clearest case.
Cross-checking your own writing. Counting the strokes you produced and comparing with the reference is a fast way to catch a stroke you have been merging or splitting, which usually indicates the order has drifted.
Reading reference material about a character, where stroke count appears alongside components and readings as part of the standard description, so knowing what it means keeps the entry readable.
Outside those, it is background information rather than a working tool, and treating it as more than that is where the planning mistakes come from.
Frequently asked questions
What is kanji stroke count?
The number of strokes needed to write a character, where a stroke is one continuous movement from the pen touching the paper to lifting it. Movements that change direction still count as one, which is why 口 is three strokes rather than four. The counting convention is standardised, so a character has one canonical number across dictionaries and databases.
Does a higher stroke count mean a kanji is harder?
No, and the belief causes a lot of unnecessary avoidance. A twenty-stroke character built from components you already know is easier than a nine-stroke character with unfamiliar structure, because you write components rather than individual strokes. Difficulty comes from unfamiliar structure, unusual proportion and similarity to other characters.
Should I learn kanji in stroke count order?
No. Stroke count correlates weakly with frequency and usefulness, so that ordering front-loads simple rare characters and postpones complex common ones. Take your order from a textbook, an exam level, the school sequence or the vocabulary you actually meet, all of which encode usefulness in a way stroke count does not.
How do I count strokes correctly?
Count each continuous movement of the pen as one stroke, including movements that turn corners, which is where most miscounts happen. If your count differs from a dictionary’s, you are usually splitting a turning stroke into two or joining two separate strokes, and that mismatch is a useful signal that your stroke order has drifted.
What is the best app for practising stroke order rather than counting?
{{appName}} checks the sequence you actually used rather than only the finished shape, prompting you first with the character hidden so the attempt is genuine production. Order is what makes characters come out consistently at speed, and it is the part a static reference cannot correct. It is free in early access.
