Japanese Numbers
Write any whole number as a Sino-Japanese numeral, with kanji, Hepburn romaji, and the euphonic sound changes handled.
12,345 written as a Sino-Japanese numeral.
Key numbers in Japanese
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 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 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 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 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.
| Number | Kanji | Romaji | Note |
|---|---|---|---|
| 1 | 一 | ichi | |
| 2 | 二 | ni | |
| 3 | 三 | san | |
| 4 | 四 | yon | shi is avoided (sounds like 死, “death”) |
| 5 | 五 | go | |
| 6 | 六 | roku | |
| 7 | 七 | nana | shichi also used |
| 8 | 八 | hachi | |
| 9 | 九 | kyū | ku also used |
| 10 | 十 | jū | not ichijū |
| 100 | 百 | hyaku | not ichihyaku |
| 300 | 三百 | sanbyaku | euphonic (not sanhyaku) |
| 600 | 六百 | roppyaku | euphonic (not rokuhyaku) |
| 800 | 八百 | happyaku | euphonic (not hachihyaku) |
| 1,000 | 千 | sen | not ichisen |
| 3,000 | 三千 | sanzen | euphonic (not sansen) |
| 8,000 | 八千 | hassen | euphonic (not hachisen) |
| 10,000 | 一万 | ichiman | always 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.