Skip to content
K Knidox Search…
Language · Numbers

Japanese Numbers

Write any whole number as a Sino-Japanese numeral, with kanji, Hepburn romaji, and the euphonic sound changes handled.

Any whole number from 0 to 99,999,999.
Try a number
In Japanese (kanji)
一万二千三百四十五

12,345 written as a Sino-Japanese numeral.

Romaji
ichiman nisen sanbyaku yonjū go
Hiragana
いちまんにせんさんびゃくよんじゅうご

Key numbers in Japanese

0 · 〇 zero
1 · 一 ichi
2 · 二 ni
3 · 三 san
4 · 四 yon
5 · 五 go
6 · 六 roku
7 · 七 nana
8 · 八 hachi
9 · 九 kyū
10 · 十 jū
100 · 百 hyaku
300 · 三百 sanbyaku
600 · 六百 roppyaku
800 · 八百 happyaku
1,000 · 千 sen
3,000 · 三千 sanzen
8,000 · 八千 hassen
10,000 · 一万 ichiman

Japanese writes 12,345 as 一万二千三百四十五, read ichiman nisen sanbyaku yonjū go. Numbers group in tens of thousands, not thousands: 万 (man) marks each 10,000. Watch the sound changes — 300 is sanbyaku, 600 roppyaku, 3000 sanzen. Enter any integer to see its kanji and romaji.

How Sino-Japanese numerals work

Japanese has two counting systems. This tool uses the Sino-Japanese set (the on’yomi readings borrowed from Chinese), which is what you use for dates, prices, phone numbers, and almost every number above ten. Each digit 1–9 has a fixed reading — ichi, ni, san, yon, go, roku, nana, hachi, kyū — and you build larger numbers by naming the power that follows each digit: 十 jū (10), 百 hyaku (100), 千 sen (1000).

The key difference from English is the grouping unit. English breaks numbers into groups of three digits (thousand, million, billion). Japanese breaks them into groups of four digits, and the big word is 万 man = 10,000. So 10,000 is 一万 (ichiman, “one man”), a million is 百万 (hyakuman, “hundred man”), and ten million is 千万 (senman, “thousand man”). Only past 100,000,000 does a new word, 億 oku, appear.

Worked example: 12,345

Split 12,345 into myriad groups: 1 (ten-thousands) and 2,345. The first group is 一万 ichiman. The second reads as 二千 (nisen) + 三百 (sanbyaku) + 四十 (yonjū) + 五 (go). Joined, 12,345 = 一万二千三百四十五, ichiman nisen sanbyaku yonjū go.

  1. 1
    Split into groups of 10,000. Read from the right in blocks of four digits, not three. 12,345 splits into 1 (× 10,000) and 2,345. Each full block is later capped with 万 (man).
  2. 2
    Read each group with 千, 百, 十. Inside a block, name each place: 2,345 → 二千 (nisen) + 三百 (sanbyaku) + 四十 (yonjū) + 五 (go). Drop empty places; a leading 1 in 10/100/1000 is silent — 100 is 百 (hyaku), not ichihyaku.
  3. 3
    Apply the euphonic sound changes. Some combinations shift for ease of speech: 300 三百 sanbyaku, 600 六百 roppyaku, 800 八百 happyaku, 3000 三千 sanzen, 8000 八千 hassen. The tool applies these automatically.
  4. 4
    Attach 万 and join. 10,000 is always 一万 (ichiman) — never bare 万. Cap the ten-thousands block with 万, then join every block: 12,345 → 一万二千三百四十五, ichiman nisen sanbyaku yonjū go.

Core numerals and the sound changes

Readings are the Sino-Japanese (on’yomi) forms in Hepburn romaji. Rows marked “euphonic” shift sound from the plain digit + power and are the ones most often misread.

NumberKanjiRomajiNote
1一ichi
2二ni
3三san
4四yonshi is avoided (sounds like 死, “death”)
5五go
6六roku
7七nanashichi also used
8八hachi
9九kyūku also used
10十jūnot ichijū
100百hyakunot ichihyaku
300三百sanbyakueuphonic (not sanhyaku)
600六百roppyakueuphonic (not rokuhyaku)
800八百happyakueuphonic (not hachihyaku)
1,000千sennot ichisen
3,000三千sanzeneuphonic (not sansen)
8,000八千hasseneuphonic (not hachisen)
10,000一万ichimanalways prefixed — never bare 万

Where learners slip: sound changes and the two systems

The euphonic changes are the whole point. A few digit + power combinations are hard to say cleanly, so the sounds fuse. With 百 (hyaku): 300 becomes sanbyaku, 600 roppyaku, and 800 happyaku. With 千 (sen): 3000 becomes sanzen and 8000 hassen. Everything else in those columns stays regular — 二百 nihyaku, 四千 yonsen — so it is only these five forms plus the silent leading “one” (十 jū, 百 hyaku, 千 sen) that you have to memorise.

Sino vs native counting. Alongside the Sino-Japanese set there is a native Japanese (kun’yomi) set that only runs 1–10: hitotsu, futatsu, mittsu, yottsu, itsutsu, muttsu, nanatsu, yattsu, kokonotsu, tō. You reach for it to count general objects (“three apples” → mittsu) and for ages of small children, but not for money, dates, or large figures. This tool writes the Sino-Japanese forms, which are the ones used for essentially every number past ten.

Why is 300 sanbyaku and not sanhyaku?
It is a euphonic (rendaku) sound change: san + hyaku fuses into sanbyaku because the b is easier to say after san. The same column gives roppyaku (600) and happyaku (800). Other hundreds stay regular, e.g. nihyaku (200) and yonhyaku (400).
How does the 万 (10,000) grouping differ from Western thousands?
Japanese counts in blocks of four digits, not three. The pivotal word is 万 (man) = 10,000. So a million is 百万 (hyakuman, “hundred man”) and ten million is 千万 (senman), rather than getting a brand-new word every three digits. A fresh word, 億 (oku), only arrives at 100,000,000.
What is the difference between the Sino and native counting systems?
The Sino-Japanese set (ichi, ni, san…) is borrowed from Chinese and is used for dates, money, and any large number. The native set (hitotsu, futatsu, mittsu…) only runs 1–10 and is used to count general objects and young children’s ages. This tool outputs the Sino-Japanese forms.
Why does 八 become happyaku and hassen?
The final sound of hachi (8) assimilates to the following consonant. Before 百 (hyaku) it gives happyaku (800); before 千 (sen) it gives hassen (8000). It is the same clipping that turns roku into roppyaku for 600.
How do I read a big number like 1,200,000?
Group it by ten-thousands: 1,200,000 is 120 × 10,000, so it is 百二十万 — hyaku nijūman. Reading the 120 block as 百 (hyaku) + 二十 (nijū) and capping it with 万 (man) gives the whole number.
Why do 4, 7, and 9 have two readings?
Four is usually yon (the reading shi is avoided because it sounds like 死, “death”); seven is nana (also shichi); nine is kyū (also ku). This tool uses yon, nana, and kyū, the readings most common when saying numbers aloud.
Why is 10,000 一万 (ichiman) and not just 万?
Unlike 十 (jū), 百 (hyaku), and 千 (sen), which drop the leading “one”, the word 万 always takes a multiplier. Ten thousand is 一万 (ichiman); leaving off the 一 would be incomplete. The same holds for 億 (oku) and higher.