The official answer is 2,136: the jōyō kanji, the list of characters designated for general use in Japan. The practical answers are smaller and depend on what you want to do, and the number you need to write by hand is smaller again than the number you need to read.
The official number
Japan’s Agency for Cultural Affairs publishes the jōyō kanji table, the list of characters for general use in official documents, newspapers and education. The current table, adopted in 2010, contains 2,136 characters.
That list is the benchmark for adult literacy in Japan, and it is a standard rather than a curriculum: it says which characters are considered general-use, not the order to learn them or how many you personally need.
Japanese schools teach a subset of these across primary school, the kyōiku kanji, spread over six years, with the remainder covered in secondary education.
Beyond the jōyō list are thousands of additional characters used in names, classical texts and specialised fields. Most learners never need them, and encountering one is a lookup rather than a gap in your knowledge.
| Set | Size | What it represents |
|---|---|---|
| Jōyō kanji | 2,136 | General-use standard for adult literacy |
| Kyōiku kanji | A subset of jōyō | Taught across primary school |
| Name kanji | Additional set | Permitted in personal names |
| Everything else | Thousands | Classical, specialist, rare |
The numbers for JLPT
If your goal is the Japanese-Language Proficiency Test, the level defines the target. The JLPT’s own level summaries describe what each level expects in terms of comprehension rather than publishing a fixed character count, and widely used study lists associate each level with an approximate set.
The important detail for a writer: the JLPT is a multiple-choice test with no handwriting section. It measures reading and listening comprehension, so preparing for it does not require producing a single character by hand.
That is worth knowing precisely because it explains a common pattern. Learners who have passed a high JLPT level and cannot write are not failing; they trained the skill the test measures, and handwriting was never part of it.
Reading, writing and typing need different numbers
Reading needs the largest set, since you meet characters chosen by other people. This is where the two-thousand-plus figures come from.
Typing needs recognition only. Japanese input converts kana to kanji and asks you to pick the right candidate, which is a recognition task, so someone can type fluently while being unable to write.
Handwriting needs the smallest set, and you choose it: the characters in the words you actually use, your own details, and whatever your studies require. A few hundred covers everyday handwriting for most learners.
Research on handwriting suggests the effort is not wasted on the larger goal: a study using brain-wave measurement found that words learned by handwriting were encoded more strongly than the same words typed, which is one reason writing a smaller set well tends to support reading a larger one.
Choosing your own number
Work backwards from the goal.
Travel and everyday interaction: a few hundred characters read, plus the ability to write your own name and address. The writing set here is tiny and specific.
JLPT: the level’s list for reading, and no handwriting requirement at all. If you want to write as well, choose a subset of the same list.
Living in Japan: the jōyō set for reading over time, plus a few hundred for handwriting, forms, notes and anything you fill in by hand. The handwriting subset here is unusually easy to identify, because it is whatever you have actually had to write since arriving, and a fortnight of noticing produces a better list than any textbook.
Reading novels or news: the jōyō set plus whatever the genre adds, which is a multi-year target and largely a reading goal.
Then split your list in two. A read list, growing through exposure, and a write list, chosen deliberately and much smaller. Trying to handwrite everything you can read is the decision that makes the review load unmanageable.
The write list is also the one that should reflect your life rather than a standard. Your own name and address, the words you use at work, the vocabulary of whatever you are actually doing in Japanese: those characters get written repeatedly in real situations, which is free reinforcement no curriculum supplies.
| Goal | Read target | Write target |
|---|---|---|
| Travel and basics | A few hundred | Name, address, a handful |
| JLPT to N3 | The level’s list | None required by the test |
| Living in Japan | Toward jōyō | A few hundred, chosen |
| Reading widely | Jōyō plus genre | Whatever you write by hand |
Why totals are a poor progress measure
Two learners claiming 1,000 kanji can be in very different positions. One recognises 1,000 in context and produces perhaps 150; the other has drilled 400 to production and recognises 1,200. The single number hides the distinction that matters.
Recognition also inflates counts, because it is what most study builds and what most self-assessment measures. That is not dishonesty, it is the gap between two abilities, and it is worth measuring separately.
A better measure for writing: pick twenty characters at random from what you have studied and write them from a prompt with nothing visible. That percentage reflects production, and it is the only number that predicts whether you can fill in a form.
What the school sequence tells you
Japanese schools spread the kyōiku subset of the jōyō list across six years of primary school, and the ordering is worth borrowing even if the timeline is not.
Early years cover characters that are frequent, structurally simple and immediately usable: numbers, days, basic nature words, common verbs. These are also heavily loaded with components that recur later, which is why the sequence accelerates rather than staying linear.
Later years add characters that are more abstract, more complex, and built from parts introduced earlier. By that stage a new character is mostly a new arrangement of known components, which is the compounding every learner eventually experiences.
Two implications for an adult learner. The school order is a reasonable default when you have no other ordering, because it was designed with frequency and structure in mind rather than assembled alphabetically. And the pace, roughly 200 characters a year for a child immersed in the language, is a useful reality check against plans that promise 2,000 in six months.
Adults can move faster than children on some of this, since they already understand what writing is for and can use structure deliberately. They also have less exposure, which is the input that consolidates everything without effort.
A sustainable pace
Three to five new written characters a day, with reviews done, is sustainable for most people. Over a year that is roughly 1,000 to 1,800, which comfortably covers everyday handwriting.
The limit is review capacity rather than learning capacity. Every new character generates future repetitions, and adding faster than you can review produces a backlog around week three, which is when most abandoned systems are abandoned.
The rate is also not constant across the whole journey. The first few hundred characters are slow because nothing is familiar, and the pace roughly doubles once components start repeating, which means an early plateau is a stage rather than a ceiling and a later acceleration is expected rather than lucky.
{{appName}} handles the production side: prompt first, character hidden, stroke order checked, misses returned sooner. It is free in early access. For pacing specifically, how many kanji to practise a day and the daily routine go into detail.
A note on the characters outside the list
Learners sometimes worry about everything beyond the jōyō set, and the worry is mostly misplaced.
Non-jōyō characters appear in three places in ordinary life. Personal and place names, which use an additional permitted set and often unusual readings; those are looked up or asked about, by native speakers too. Older texts and literary writing, where less common characters and pre-reform forms appear. And specialised fields, where a particular trade or discipline carries its own small vocabulary.
None of those is a gap in general literacy. A Japanese adult meeting an unfamiliar character in a name does the same thing a learner does, which is ask or look it up, and that is worth knowing because it recalibrates what fluency means: not knowing every character, but knowing the general-use set and being comfortable looking up the rest.
The practical consequence for study is to stop treating rare characters as a target. Time spent on the general-use set and on the vocabulary that uses it returns far more than time spent extending a count into territory you will meet twice a year.
Measuring what you actually have
A number is only useful if it corresponds to something you can do, so it is worth defining three separately.
Passive recognition: you see the character in a word, in context, and understand the word. This is the largest number and the one most study builds.
Isolated recognition: you see the character alone and can give a meaning and a reading. Smaller, and a better proxy for whether the character is really known rather than guessed from context.
Production: you get a prompt and write the character with nothing visible. Smallest by a wide margin, and the only one that predicts whether you can fill in a form or write a note by hand.
Testing all three takes ten minutes: twenty characters in context, twenty in isolation, twenty from a prompt. The three numbers together tell you what to do next far better than any single total, and the gap between the first and the third is usually the surprise that changes how someone studies.
Frequently asked questions
How many kanji do you need to know?
The official general-use standard is the jōyō list of 2,136 characters, which is the benchmark for adult literacy in Japan. In practice the number depends on your goal: a few hundred for travel and basics, the relevant list for a JLPT level, and toward the jōyō set for living in Japan or reading widely. Handwriting needs far fewer than reading.
How many kanji are there in total?
Tens of thousands exist in historical dictionaries, and the number in practical use is far smaller. The jōyō list of 2,136 covers general use, an additional set is permitted for personal names, and beyond that lie characters used in classical texts and specialised fields. Meeting one of those is a lookup rather than a gap in your knowledge.
How many kanji do you need for the JLPT?
Each level has an approximate associated set, with the official level summaries describing expected comprehension rather than publishing fixed counts. The detail that matters for writers: the JLPT is multiple choice with no handwriting section, so passing any level requires recognition rather than the ability to produce characters by hand.
How many kanji should I be able to write by hand?
Fewer than you can read, and the ones you actually use: the characters in your active vocabulary, your own details, and whatever your studies demand. A few hundred covers everyday handwriting for most learners. Growing that at three to five new characters a day alongside reviews is sustainable; handwriting everything you can read is not.
What is the best app for building a writable kanji set?
{{appName}} measures production rather than exposure: it prompts you first, keeps the character hidden until you have written it, checks your stroke order and returns missed characters sooner. That makes its count the useful one, since it reflects what you can produce cold rather than what you have seen. It is free in early access.
