Affiliate disclosure: some links in my articles are affiliate links, which means I may earn a commission if you buy through one, at no extra cost to you. As of 4 September 2026 I have four approved affiliate partnerships. One of them is JapanesePod101, which is one of the products I assess on this page, and it pays 25 percent on a cookie that does not expire. Two of the others are Japanese knife retailers and have nothing to do with language learning. The fourth is The Japan Shop, a bookshop approved on 4 September 2026, and that one sells JLPT material of exactly the kind this page is about. On 4 September 2026, after the approval, I added one section to this page, and it holds the only tracked link here: a 28 dollar digital bundle of N5 exam material that pays me 50 percent, about 14 dollars. That section carries the date it was written so you can tell it apart with certainty. Everything else below was written before any of these approvals and has not been moved, reworded or re-ranked since. My affiliate disclosure page keeps the current list of every tracked link on this site. You should know before you start that the single most valuable study resource on this page is free and published by the people who write the exam, and that the product I rank first for structure pays me nothing at all.
I have never sat the Japanese Language Proficiency Test. I was never going to. So I want to be honest about what I can and cannot tell you here.
I cannot tell you which app got somebody through N2 in eight months. I have no exam technique to pass on, and the people who do have it are easy to find.
What I can tell you is what the test measures against the language as it is actually used, because I am the thing it is measuring against. And once I read the official specification carefully, something clicked into place that explains every single article I have written on this site.
The JLPT does not test speaking. It does not test writing. There is no production component of any kind, at any level, from N5 to N1. Every question is multiple choice. The official site describes the competence being measured in terms of language activities, and it names two: Reading and Listening.
The same holds for the business standard, which I did not expect. The BJT business Japanese test is also listening and reading only, in four-option multiple choice, and a score of 480 is worth 15 immigration points.
Hold that next to the rest of this site for a moment. I have written that no product makes you produce a correct pitch accent. That none of them makes you choose a politeness level for a specific person. That none of them puts you in the listening half of a conversation and requires a response at the right moment.
I kept describing these as gaps in the products. They are not. They are gaps in the exam, and the products are aligned to the exam. Nobody built a pitch accent drill because nothing in the industry rewards one. The test is the reason, and until now I had been blaming the wrong layer.
So this article does two things. It tells you which apps to use for the JLPT, honestly and by level. And it tells you exactly what the certificate at the end will and will not mean, from the side of the desk you are hoping to sit across from.
What the test actually is
Straight from the official specification, so we are working from the same facts.
| Level | Sections | Time |
|---|---|---|
| N1 | Language Knowledge with Reading; Listening | 110 min + 55 min |
| N2 | Language Knowledge with Reading; Listening | 105 min + 50 min |
| N3 | Vocabulary; Grammar with Reading; Listening | 30 + 70 + 40 min |
| N4 | Vocabulary; Grammar with Reading; Listening | 25 + 55 + 35 min |
| N5 | Vocabulary; Grammar with Reading; Listening | 20 + 40 + 30 min |
Three things follow from that table, and all three matter more than any app recommendation I can give you.
First, you never generate a Japanese sentence. Not one, in any level, at any point. You select from options that somebody else wrote.
Second, the listening section makes you a spectator. You hear a conversation and answer questions about it afterwards. You are never a participant, never on a clock, never required to react while somebody is still talking. The skill being measured is comprehension of recorded speech, which is a real skill and is not the same skill as holding a conversation.
Third, the whole thing is recognition. Recognition is genuinely easier than production, in every language, for every learner. You can reliably recognise several times more vocabulary than you can retrieve under pressure. A test built entirely on recognition is measuring the larger and softer of the two numbers.
None of this makes the JLPT a bad test. It makes it a specific test, and one that is honest about its scope on its own website. The problem is what happens downstream, when it becomes the only measurement anybody has.
Why every learning product looks the way it does
Here is the mechanism, and I think it is the most useful thing on this page.
The JLPT is the only widely recognised external standard for Japanese. Employers ask for it. Visa point systems count it. Universities require it. It is the number that goes on a form.
So it becomes the target. Products organise themselves by N5 to N1 because that is what learners search for. Learners study toward the levels because that is what employers ask for. And every skill outside the exam specification quietly loses its funding, because no product can charge for something that does not show up on the only scoreboard that exists.
Counters are a small and clean example. The exam tests them by recognition only, and so do the products, which I traced through in my comparison of Japanese counter apps.
The exam has no writing section either, which is why composition feedback is so thin a market. I go through it in my guide to Japanese writing practice apps.
Look at what falls outside the specification:
- Pitch accent. Not tested. Not in any section, at any level.
- Choosing a register for a specific person. Not tested. You may be asked to recognise which form is polite, which is a different question.
- Backchannelling, the listener half of a conversation. Structurally untestable in the format, because you never respond to anything.
- Producing keigo with the direction correct. Recognition of honorific forms appears at N2 and N1, and I go into what that leaves out in the keigo article. Deciding whose action it is, live, does not.
- Speaking at all. Not tested.
That is not a coincidental list. That is the same list I have been compiling article by article for two months, arrived at from the opposite direction. I was cataloguing what products fail to teach. The exam specification predicts the whole catalogue in advance.
What the certificate means to a Japanese employer
I want to be useful rather than cynical here, because both the enthusiasm and the dismissal you will read online are overdone.
N1 and N2 are real. They are asked for by name in Japanese job listings, they clear an HR filter that is otherwise very hard to clear from outside the country, and passing N1 is a substantial achievement that no amount of arguing about test design takes away.
And everyone on the hiring side knows what they do not guarantee. The phrase you will hear in Japanese offices is that somebody has the certificate but cannot hold a meeting. This is not a slur invented by learners on forums. It is a specific, recognised outcome of a recognition test, and hiring managers compensate for it by adding an interview in Japanese, which tests the thing the certificate did not.
So the practical read is this. The certificate opens the door and the interview decides. Study for the test, get the number, and understand that on the day you use it, you will be assessed for forty minutes on precisely the skills the test declined to measure.
Which means the smartest possible JLPT strategy is to prepare for the exam efficiently, so that you have time left over for the things the exam ignores. Efficiency in test prep is not laziness here. It is what buys you the hours to work on speaking.
How I judged these
Three questions.
1. Is it actually aligned to the exam? Does it follow the N5 to N1 structure, or is it a good product that happens to teach Japanese in a different order?
2. Does it cover the level you are sitting? Several well-known products stop early. This is the single most common way learners waste money on JLPT prep.
3. Does it leave you time for what the exam does not test? A product that eats twenty hours a week teaching you to recognise vocabulary has taken those hours from somewhere.
The comparison
| Resource | Aligned to the exam? | Levels covered | Cost | Can it pay me? |
|---|---|---|---|---|
| Official sample questions and workbooks | Perfectly. Written by the examiners | N1 to N5 | Samples are free | No. Nothing |
| Bunpro | Yes. Grammar is organised by JLPT level | N5 to N1 | Low monthly or lifetime | No — points, not cash |
| JapanesePod101 | Yes. Dedicated JLPT pathways and courses | Strongest at N5 to N3 | Subscription | Yes |
| Anki with a shared deck | As aligned as the deck you choose | Any, deck dependent | Free | No — free and open source |
| The Japan Shop JLPT N5 digital bundle | Yes. Built book by book on the exam sections | N5 only. An N4 bundle exists, and nothing digital above it | 28 dollars, once | Yes — 50 percent, about 14 dollars to me |
| Dedicated JLPT quiz apps | Yes, and narrowly so | N5 to N1 | Free or ad supported | No |
| WaniKani | No. It follows its own kanji order | Not applicable | Monthly | No — no affiliate program exists |
| Rocket Japanese | Partly. Official: Level 1 is N5, Levels 2 and 3 make N4 | Stops at N4 | One time purchase | Yes, 40 percent — the second highest rate here, and I am not in it |
| Pimsleur | No. Audio only, so no reading at all | Not applicable | Subscription | Yes, and the most per sale |
| A tutor on italki | Only if you ask | Any | Hourly | About ten dollars |
Resource by resource
The official sample questions — start here, and they cost nothing
The organisation that writes the exam publishes sample questions for every level from N1 to N5, free, on the official site, along with paid official practice workbooks.
I am putting this first because the single largest avoidable mistake in JLPT preparation is not knowing the shape of the questions. The format is idiosyncratic. Sentence ordering problems, where you assemble a scrambled sentence and report which fragment lands in the starred position, are not a thing you will meet anywhere else in life. Meeting that question type for the first time in the exam hall is a self inflicted wound.
Work through the official samples before you buy anything. They will also tell you honestly which level you should be sitting, which is worth more than any recommendation from me. Can it pay me? Nothing.
Bunpro — the closest thing to a JLPT machine, and it pays me nothing
Bunpro organises grammar by JLPT level, from N5 through N1, and drills it with spaced repetition where you type the answer. For grammar, which is a large slice of the Language Knowledge section, this is the most directly aligned tool I know of that is not published by the examiners. The exam is also the reason grammar is sold as a numbered list of points at all, which I take apart in the grammar app comparison.
It is also the product I have now recommended in four consecutive articles for a reason that has nothing to do with the JLPT: it is the only one in this whole category that makes you produce rather than recognise. That property is a slight mismatch with a recognition exam and an excellent match with the rest of your life.
It has no cash affiliate program. It never has. Can it pay me? No.
JapanesePod101 — real JLPT pathways, strongest at the lower levels
There are dedicated JLPT sections in the lesson library, including full JLPT seasons and curated N5 and N4 pathways that strip the catalogue down to exam relevant material. The listening focus is a genuine advantage here, because the listening section is where learners who studied only from text get ambushed.
The honest limitation is depth at the top. The library is deepest where most learners are, which is N5 to N3, and if you are sitting N1 you will outgrow it. Can it pay me? Yes.
Anki — free, and only as good as the deck
There are mature shared decks for every JLPT level. Anki costs nothing on desktop and Android, it is open source, and for raw vocabulary volume it is as effective as anything paid.
The cost is setup and discipline. You are choosing the deck, and a badly built deck will drill you on words the exam does not use. If you enjoy tinkering, this is the highest value per yen on the entire page. If you do not, you will build a beautiful configuration and stop studying. Can it pay me? No.
The Japan Shop JLPT N5 bundle — the one thing on this page that pays me
Added 4 September 2026, after the approval described at the top of this article. This is the only part of this page written after that date, and it holds the only tracked link on the page. Nothing above or below it was moved, reworded or re-ranked to make room for it. The free official samples are still first, and they are still first for the same reason they always were.
Seven digital books, as the shop counts them, aimed at a single exam level and sold together for 28 dollars as a one time purchase: vocabulary, grammar, reading, a kanji book covering the 103 characters the publisher lists for N5, sentence patterns, and kanji reading practice. More than 2,000 audio files, Anki decks built to match the books, and a 45 page practice test with an answer key.
Why it sits here, directly after Anki. Read the Anki entry above again. The weakness I named there is that you are choosing the deck yourself, and a badly built deck will drill you on words the exam does not use. This is the same idea with that specific risk taken out: the decks arrive already matched to one published exam level, alongside the books they came out of. If you enjoy assembling your own stack, the free route above is better value and I mean that. If choosing a deck is the step that has stalled you twice, 28 dollars buys the decision.
The ceiling, stated as plainly as I state it for Rocket three sections below. This bundle is N5 only. There is a separate N4 bundle at 19 dollars, and I checked the whole catalogue on 4 September 2026: the shop publishes nothing digital above N4. So the criticism I make of Rocket applies here too, and this time it applies to the thing that pays me. If you are sitting N3, N2 or N1, this is not for you, and no commission is going to make me write otherwise.
What I have not done. I have not worked through these books. What I checked on 4 September 2026 is what is verifiable: the count of seven books, the formats, the audio, the Anki decks, the practice test, the level and the price. I cannot tell you whether the explanations are well written, and I am not buying the bundle in order to find out. Weigh that as a limit on this recommendation: everything above rests on what the shop publishes and on what I could verify from outside, and none of it rests on my having worked a single page.
One thing to know before you click. The same shop sells a 49 dollar bundle that contains this one plus a beginner course, and it would pay me nearly twice as much. I am not pointing you at that one. Most of what it adds is kana and absolute beginner material, and if you are shopping for JLPT preparation you already have that.
Can it pay me? Yes. 50 percent on digital bundles, so about 14 dollars. That is the only payment available to me anywhere on this page, which is exactly why the free official samples above are the recommendation I would defend hardest. The JLPT N5 digital bundle at The Japan Shop. You never pay more through that link.
The dedicated JLPT quiz apps — high volume, low nutrition
Search either app store for the JLPT and you will find apps advertising tens of thousands of practice questions and dozens of past exams. They rank well and several are free or ad supported.
I have not installed these, so I am not reviewing their quality, and I would want to know where a free app claiming forty thousand past questions obtained them. What I will say is structural: mock question grinding is a legitimate part of test preparation, and it is the part with the sharpest diminishing returns. Once you know the question formats and can complete a section in time, further mock questions mostly measure what you already know. Can it pay me? Nothing.
WaniKani — excellent, and not a JLPT tool
WaniKani teaches kanji and vocabulary in its own carefully designed order, built around radicals and mnemonics rather than exam levels. That order is deliberately not the JLPT order, which is a recurring topic in the community, and a fairly common kanji can sit at a surprisingly high WaniKani level.
If you have six months until N3, WaniKani is the wrong tool, and I say that as somebody who recommends it more often than anything else for learners without a deadline. It is a long game product being asked to do a short game job. Can it pay me? No, and no program exists.
Rocket Japanese — check the ceiling before you buy
Rocket states its own coverage plainly, which I appreciate: Level 1 is equivalent to N5, and Levels 2 and 3 together give you the grammar and vocabulary for N4.
Read that again if you are aiming higher, because it is the practical point. Rocket does not cover N3, N2 or N1. For a beginner sitting N5 or N4 it is a genuinely structured course with strong audio. For anybody past that, the money is misallocated.
This is the fourth article in which I have qualified or marked down Rocket, and at 40 percent it pays the second highest commission rate of anything I could link on this page, behind only the digital bundle above. I would rather tell you where the course stops than sell you a level it does not reach. Can it pay me? Yes, at 40 percent, the second highest rate here — I am not in that programme.
Pimsleur — structurally unable to help you here
Pimsleur is audio. There is no reading component to speak of, and reading plus vocabulary and grammar in written form is the larger half of every JLPT sitting.
I rate the Pimsleur method highly for speaking, and I have said so in the main comparison. For this specific goal it cannot do the job, and it carries the largest single commission of anything I have looked at in this cluster. Can it pay me? Yes, the most per sale, and it is still the wrong purchase for this.
A tutor — not for the test, for what comes after it
Tutoring is inefficient JLPT preparation. An hour of conversation practice will move your score less than an hour of grammar drilling, and I am not going to pretend otherwise for a ten dollar referral, which I am not signed up for in any case.
Book one anyway, at a low frequency, for the reason set out at the top of this article. The exam does not test production, the interview will, and the gap between those two is where people who passed N1 discover they cannot say what they know. Half an hour a week alongside your drilling keeps the production side alive. Can it pay me? About ten dollars.
What the listening section sounds like to a native ear
I listened to official listening samples across several levels, and I want to describe them accurately, because learners tend to hold two wrong beliefs about this section at the same time.
The first wrong belief is that JLPT listening is real Japanese. It is not, and it does not claim to be. It is studio Japanese: two speakers, clean turns, no overlap, no interruption, every sentence finished. The delivery is clearer than a real conversation in a way that has nothing to do with speed. Real speech is not primarily hard because it is fast. It is hard because people trail off, talk over each other, and leave half of every sentence for you to infer.
The second wrong belief is that this makes the section easy, and that is where people lose marks. The audio is played once. There is no replay, you cannot go back, and while you are still deciding on question four the recording has moved to question five. What the section really tests is not comprehension so much as comprehension without hesitation, which is a different and more brutal thing. A single unfamiliar word can cost you the next question as well, because you are still processing it when the audio moves on.
So the preparation that works is not more listening in general. It is timed listening where you are forbidden to rewind, followed by checking the transcript afterwards. Anything with a replay button trains a habit the exam will punish.
And there is a corollary worth stating plainly. Passing the listening section does not mean you can follow a conversation you are part of. In the exam you are a spectator with a defined question. In a real conversation you are a participant with an obligation to respond, and I have argued elsewhere that the listener in Japanese carries a continuous audible job that no exam has ever tested. These are different skills that share a name.
A study stack by level
N5 and N4. JapanesePod101 pathways or Rocket for structure, Bunpro for grammar, Anki or WaniKani for kanji if you have time, and the official samples about a month before the exam. Rocket is fully in scope at these levels and nowhere else.
N3. Bunpro becomes the centre of gravity. JapanesePod101 still has real depth here. Add graded readers or news written for learners, because reading speed starts to be the thing that fails people rather than knowledge. My guide to reading Japanese covers the tooling for exactly this stage.
N2 and N1. Bunpro plus volume reading plus the official workbooks. Products thin out sharply at this altitude and the honest answer is that you move to native material earlier than you feel ready to. At N1 the exam is testing whether you can process real Japanese at speed, so the preparation converges on doing exactly that.
At every level. Do timed practice with the official samples early rather than late. Running out of time is the most common way to fail a section you actually knew.
What I would want you to take from this
The JLPT is a good test of the things it tests, and it is the only door of its kind, so walk through it. Prepare for it efficiently and without romance. It is a recognition exam, it rewards drilling, and drilling is a solved problem.
But do not let the exam define what Japanese is. Everything this site has documented as missing from learning products is missing because it is missing from the test. Pitch, register, the listener half of a conversation, honorifics with the direction correct: none of it is examinable in a multiple choice format, so none of it is built, so none of it gets learned, and the outcome is a category of learner who holds a certificate and freezes in a meeting.
You can see that outcome coming from where you are standing right now. The fix is not to skip the test. It is to keep half an hour a week for the part of the language nobody is going to grade.
One last thing, from my side of it
I said at the top that I have never taken this test, and I want to close on what happened when I looked at the N1 sample questions out of curiosity.
I got them right. I also could not have explained why several of them were right. The grammar section asks you to distinguish between constructions that I use correctly every day and have never once thought about as objects, and a couple of the questions were asking about distinctions I would have to reverse engineer from my own speech to describe.
That is not a boast about being a native speaker and it is not a complaint about the test. It is the same thing I have found in every article here: producing this language and describing it are two different skills, and you are being asked to do the harder one first. You are learning the map before the territory, in an order no native has ever experienced.
So when the exam feels unnaturally analytical, that is because it is. It is the only way to test a language on paper. Pass it, take the certificate, and then go and get the territory, which is where the rest of this site lives.