The commonly cited cumulative figures:
| Level | Vocabulary (estimated) |
|---|---|
| N5 | ~800 |
| N4 | ~1,500 |
| N3 | ~3,700 |
| N2 | ~6,000 |
| N1 | ~10,000 |
Read these as estimates
The JLPT has published no official vocabulary list since 2010. The redesign removed the published specification on purpose, so that the test measures the ability to use Japanese rather than the ability to memorise a defined list.
Everything circulating online is a reconstruction. Useful for planning; not authoritative. A site describing its list as official is misrepresenting it.
Where the real jumps are
Look at the gaps rather than the totals:
- N5 to N4 — roughly 700 more words. Manageable.
- N4 to N3 — roughly 2,200 more. This is the big one, and it is why N4 to N3 feels so much harder than the step before it.
- N3 to N2 — roughly 2,300 more, and the character of the vocabulary changes.
- N2 to N1 — roughly 4,000 more, much of it abstract and specialised.
Vocabulary is the bottleneck at every level above N4. Not grammar. Grammar at N2 fits in a book you could work through in a few months; the vocabulary takes a year.
The change at N2 that catches people
N3 vocabulary is concrete: objects, actions, places, times. You meet it constantly if you live in Japan, so immersion does much of the work.
N2 vocabulary is abstract: opinion, cause and effect, analysis, administration, business. You meet it in writing far more than in conversation, which is why living in Japan helps less with N2 than it does with N3, and why reading becomes non-negotiable at that level.
How to learn them
Spaced repetition, daily, forever. Twenty a day at N5 and N4; thirty at N3 and above. The number that matters is not how many you add but how many you retain, and retention is a function of daily review.
In sentences, not as translation pairs. A word learned inside a sentence carries its particles, its register and its collocations. A word learned as “X = Y” produces recognition without usability.
With the confusable neighbour attached. The exam’s wrong answers are built from near-misses, so learn near-misses together.
Through volume of reading, which is what moves a word from “I know this” to “I did not notice I knew this”. Spaced repetition builds the inventory; reading makes it automatic. You need both.
The thing the numbers hide
A word is not one item. It has a reading, a meaning, a register, and a set of things it habitually appears with. “6,000 words at N2” understates the work considerably.
Which is the argument for starting earlier and going slower rather than trying to compress it. There is no version of N2 or N1 that does not involve a year or more of daily vocabulary work, and plans that promise otherwise are describing something that does not happen.
For JFT-Basic
A much smaller and more specific target: roughly 1,500 to 2,500 words, weighted hard towards daily life and the workplace. See how many words you need for JFT-Basic.