Japanese numbers look like a one afternoon job. Ten words, a pattern for the teens, a pattern for the tens, done. Then you try to say your phone number and something goes wrong, and you find out that four has two readings, that one of them is avoided because it sounds like death, that three hundred is not san-hyaku, and that Japanese stops grouping digits in threes at exactly the point English speakers need it most.
None of that is hard. It is just never all in one place. So here it is: the counting you need, the irregular parts flagged where they actually bite, and the big number system that trips up people who have been studying for years.
Table of Contents
- Japanese has two counting systems
- Numbers 1 to 10
- The three numbers with two readings
- Numbers 11 to 99
- Hundreds, and the three that change sound
- Thousands
- Ten thousand: where big numbers stop making sense
- How to say zero
- Numbers in real situations
- What to practice first
Japanese has two counting systems
This is the part that explains most of the confusion, so it goes first.
Japanese has two sets of number words. They are not equal partners, and the difference in scope is the point:
The Sino-Japanese set (一, 二, 三) came from Chinese. It runs from one to infinity, and it is what you use with counter words, dates, money, times, phone numbers and arithmetic. This is the set that matters most, and it is the one this article mostly covers.
The native Japanese set (一つ, 二つ, 三つ) is older, and it stops at ten. You use it to count generic objects when you do not know or do not need a specific counter word.
In practice: if you are counting a thing and you know its counter, use the Sino-Japanese set plus that counter. If you are pointing at some apples and want three, say 三つ and you will always be understood. Beginners can lean on the native set as a safety net, which is exactly what it is there for.
Numbers 1 to 10
| # | Kanji | Sino-Japanese | Native Japanese |
|---|---|---|---|
| 1 | 一 | いち (ichi) | ひとつ (hitotsu) |
| 2 | 二 | に (ni) | ふたつ (futatsu) |
| 3 | 三 | さん (san) | みっつ (mittsu) |
| 4 | 四 | よん / し (yon / shi) | よっつ (yottsu) |
| 5 | 五 | ご (go) | いつつ (itsutsu) |
| 6 | 六 | ろく (roku) | むっつ (muttsu) |
| 7 | 七 | なな / しち (nana / shichi) | ななつ (nanatsu) |
| 8 | 八 | はち (hachi) | やっつ (yattsu) |
| 9 | 九 | きゅう / く (kyuu / ku) | ここのつ (kokonotsu) |
| 10 | 十 | じゅう (juu) | とお (tou) |
The native column ends at ten and has no eleven. After that everyone uses the Sino-Japanese set.
The three numbers with two readings
Four, seven and nine each have two readings, and this is the single most common source of Japanese number mistakes. The rule of thumb is easy, and then there are fixed expressions that ignore it.
Default to よん (yon), なな (nana) and きゅう (kyuu). If you are counting, giving a phone number, saying a price, or unsure, these are safe and natural.
Why し (shi) gets avoided: it is a homophone of 死, death. The association is real enough that よん became the everyday default, and you will sometimes see a building skip a fourth floor or room number, though how consistently that happens varies a lot. A similar preference pushes seven toward なな.
But some expressions are frozen, and there the "avoided" reading is the only correct one:
| Expression | Reading | Not |
|---|---|---|
| 四月 (April) | shigatsu | |
| 七月 (July) | shichigatsu | |
| 九月 (September) | kugatsu | |
| 四時 (4 o'clock) | yoji | |
| 七時 (7 o'clock) | shichiji | |
| 九時 (9 o'clock) | kuji |
Four o'clock is worth staring at: it is neither よん nor し, it is よ. Months and clock times are the two places these irregulars cluster, and there is more of this in the guide on how to tell time in Japanese.
Numbers 11 to 99
Here the language pays you back. There are no new words to learn, only two patterns.
Eleven to nineteen: ten, then the digit.
- 11 = 十一 (juuichi), literally "ten one"
- 15 = 十五 (juugo)
- 19 = 十九 (juukyuu)
Twenty to ninety: the digit, then ten.
- 20 = 二十 (nijuu), literally "two ten"
- 50 = 五十 (gojuu)
- 90 = 九十 (kyuujuu)
Then combine both. Anything up to 99 is just tens plus ones, in that order, with nothing joining them:
- 42 = 四十二 (yonjuuni)
- 68 = 六十八 (rokujuuhachi)
- 99 = 九十九 (kyuujuukyuu)
No sound changes anywhere in this range. If you can say one to ten, you can say one to ninety-nine today.
Hundreds, and the three that change sound
Hundred is 百 (hyaku). Note that 100 on its own is just ひゃく, never ichihyaku.
Three of the hundreds change sound, because san-hyaku, roku-hyaku and hachi-hyaku are awkward to say quickly. Japanese solved this the way it usually does, by softening or doubling the consonant.
| # | Japanese | Romaji |
|---|---|---|
| 100 | ひゃく | hyaku |
| 200 | にひゃく | nihyaku |
| 300 | さんびゃく | sanbyaku |
| 400 | よんひゃく | yonhyaku |
| 500 | ごひゃく | gohyaku |
| 600 | ろっぴゃく | roppyaku |
| 700 | ななひゃく | nanahyaku |
| 800 | はっぴゃく | happyaku |
| 900 | きゅうひゃく | kyuuhyaku |
Three, six and eight. If you remember only that those are the odd ones, you will catch yourself before you say it wrong.
Thousands
Thousand is 千 (sen), and it behaves almost the same way. Again, 1,000 alone is just せん, not ichisen.
| # | Japanese | Romaji |
|---|---|---|
| 1,000 | せん | sen |
| 2,000 | にせん | nisen |
| 3,000 | さんぜん | sanzen |
| 4,000 | よんせん | yonsen |
| 5,000 | ごせん | gosen |
| 6,000 | ろくせん | rokusen |
| 7,000 | ななせん | nanasen |
| 8,000 | はっせん | hassen |
| 9,000 | きゅうせん | kyuusen |
Only two irregulars this time, three and eight. Six is regular here even though it changed in the hundreds, which is annoying but is genuinely the whole rule.
Ten thousand: where big numbers stop making sense
This is the part that catches out learners who are otherwise fluent, and it is not a vocabulary problem. It is a formatting problem.
English groups digits in threes: thousand, million, billion. Japanese groups them in fours. The unit is 万 (man), which is ten thousand, and once you pass it every English instinct you have points the wrong way.
| Figure | Japanese | Romaji | In English |
|---|---|---|---|
| 10,000 | 一万 | ichiman | ten thousand |
| 50,000 | 五万 | goman | fifty thousand |
| 100,000 | 十万 | juuman | one hundred thousand |
| 1,000,000 | 百万 | hyakuman | one million |
| 10,000,000 | 千万 | senman | ten million |
| 100,000,000 | 一億 | ichioku | one hundred million |
Read that table again slowly. A million in Japanese is "hundred ten-thousand". There is no single word for million, and there is no unit at all between 万 and 億 (oku, a hundred million).
Note also that 万 requires its one: ten thousand is いちまん, never just まん. Hundred and thousand drop it, ten thousand does not.
The practical trick: when you see a large number written with English commas, ignore them and re-split it from the right in groups of four. 2,500,000 becomes 250 | 0000, which is 250 万, にひゃくごじゅうまん. Do the regrouping on paper a few times and it becomes automatic. Try to translate unit by unit instead and you will be stuck forever.
This matters more than it sounds, because Japanese salaries, house prices and news figures are all quoted in 万.
How to say zero
Three options, and they are not interchangeable.
- ゼロ (zero) is the everyday default, borrowed from English and written in katakana. Use it for most things.
- 〇 (maru) literally means "circle" and is used for a zero sitting inside a sequence, such as a room number or a formal read-out.
- 零 (rei) is formal and written. You will hear it in weather reports and see it on official documents.
If ゼロ is katakana to you right now, the hiragana and katakana quiz is the fastest way to fix that, since numbers show up constantly in beginner reading.
Numbers in real situations
Knowing the words is not the same as using them. Four places where they behave slightly differently:
Phone numbers are read digit by digit, never as whole numbers, and the hyphen is read as の. Four, seven and nine usually take よん, なな and きゅう, but this is a tendency rather than a rule: readings vary between speakers, and seven still turns up as しち. Zero is most often ゼロ, with まる common when reading out a sequence and れい in formal or official contexts. So 03-1234-5678 would typically be read ゼロさん の いちにさんよん の ごろくななはち.
Money uses 円 (en) attached directly to the number: 100 yen is 百円 (hyakuen), 3,000 yen is 三千円 (sanzen'en). The sound changes above all still apply.
Counting objects needs a counter word, and that is a whole system of its own: different words for flat things, long things, machines, people and animals. It is the natural next step after this article, and it is covered in the Japanese counter words guide. You can drill them in the counter practice tool.
Dates are the least regular part of the language's number system. The 1st through the 10th barely follow the pattern at all: the 1st is 一日 (tsuitachi), the 2nd is 二日 (futsuka), the 3rd is 三日 (mikka), and the 10th is 十日 (tooka). Most days above ten are regular, but three more stay irregular and are the ones that catch people out: the 14th is 十四日 (juuyokka), the 20th is 二十日 (hatsuka), and the 24th is 二十四日 (nijuuyokka). That is thirteen days to memorise rather than derive, and the rest behave.
What to practice first
In order, because the order genuinely matters:
- One to ten in the Sino-Japanese set, with よん, なな and きゅう as your defaults. This is the foundation for everything else.
- Eleven to ninety-nine. Pure pattern, no new words, and it makes you feel fast.
- The six sound changes: 300, 600, 800, 3,000, 8,000, and the fact that 100, 1,000 drop their "one" but 10,000 does not.
- The 万 regrouping. Practise reading prices, because that is where you will meet it.
- Counters, once the numbers themselves are automatic.
The native set (ひとつ, ふたつ) is worth learning early too, precisely because it rescues you when you have forgotten which counter a thing takes.
If you want to see numbers inside real sentences rather than in tables, paste one into the Japanese sentence analyzer and it will break the number, the counter and the particle apart for you. There is also a numbers vocabulary list if you just want the words on one page, and the JLPT N5 grammar list for the sentence patterns you will be dropping these numbers into.

