How the test works
The test uses a list of the 10,000 most frequent Japanese words, split into five bands of 2,000 by how common they are. It picks 20 words at random from each band and mixes them together, so you see everyday words like 食べる alongside rarer ones. Each test draws a fresh set of words, which means two attempts will not show you the same questions.
For each word you say whether you know it. A no moves straight on. A yes is followed by a meaning check: choose the right English meaning from four options. Only words you match to their meaning count towards your score, so a confident guess about a familiar-looking word does not inflate the result. If you are torn between two meanings, choose Not sure rather than guessing.
Reading your result
Your score in each band is scaled up to the band's 2,000 words and added together. If you matched 15 of the 20 most common words, the test assumes you know about three quarters of that band. The range shows how far the estimate could move from the luck of which words were picked; the bars show where your knowledge starts to thin out.
Treat the number as a rough guide, not a measurement. It has not been checked against learners' scores on other tests. A lucky guess on the meaning check still counts. A word you know in a different sense from the one shown may count as a miss, and people differ in what they are willing to call known. The test only covers the 10,000 most common words, so a result near 10,000 means you have outgrown it rather than that your vocabulary stops there.
What the number is useful for
A rough count is a good way to see progress over months, or to compare yourself before and after a stretch of reading. Taking the test again in a few weeks gives you a different set of words, so your result is not inflated by remembering last time's answers. The band bars are often more useful than the total: if your scores drop off sharply after the first band, the next most valuable step is to fill in common everyday vocabulary rather than chasing rare words.
The estimate is not a JLPT level. Exam levels test grammar, reading speed and listening as well as vocabulary, and their word lists are not published. Word counts often quoted for each level are community estimates rather than official targets.
Where the words come from
The frequency ranking comes from Tatoeba's sentence collection, and the meanings from the JMdict dictionary. When two common words share a spelling, the test shows the reading or a kanji spelling to tell them apart. The wrong answers are taken from other words of a similar kind and frequency, so a verb's choices are all verbs, and choices that share a meaning with the right answer are left out. Full credits are on the tool data credits page.