Onyomi vs Kunyomi: How to Know Which Reading to Use

Pick up almost any kanji and you’ll find it has not one but two or more pronunciations — and which one applies depends entirely on the word it’s sitting in. This is the onyomi vs kunyomi distinction, and while it sounds like one more thing to memorise, it actually follows a pattern reliable enough to guess correctly most of the time. Here’s what the two reading types are, why they exist, and the rule of thumb that gets you there fast.

This builds on our how to learn kanji guide — if you’re still building your kanji foundation, start there first.

Why one kanji has multiple readings

The short version: Japan didn’t invent kanji, it borrowed them.

Roughly 1,500 years ago, Chinese characters arrived in Japan along with their Chinese pronunciations. But spoken Japanese already existed, with its own native words for the same concepts. Rather than pick one system, Japan kept both:

• The native Japanese word got attached to the kanji that matched its meaning — this became the kunyomi.
• The borrowed Chinese pronunciation stuck around for use in words built the Chinese way — this became the onyomi.

The result: a single character carrying two separate pronunciation traditions, layered on top of each other for over a millennium. That’s why 食 can be read ta (in 食べる) or shoku (in 食事) depending entirely on context.

Onyomi (音読み) — the borrowed reading

Onyomi, literally “sound reading,” is the Japanese approximation of how the character was pronounced in Chinese when it was borrowed.

Onyomi readings show up almost exclusively in compound words (熟語, jukugo) — two or more kanji stacked directly together, no kana in between:

食事 (shoku-ji) — “meal”

学校 (gaku-kou) — “school”

会話 (kai-wa) — “conversation”

Notice the pattern: no hiragana breaking up the kanji. Onyomi readings are typically short (often one or two syllables) and tend to sound more “clipped” than kunyomi.

Kunyomi (訓読み) — the native reading

Kunyomi, literally “meaning reading,” is the original Japanese word that existed before kanji arrived, mapped onto the character that matches its meaning.

Kunyomi shows up when a kanji stands alone, and especially when it’s followed by okurigana — trailing hiragana that completes the word and carries conjugation:

食べる (ta-beru) — “to eat”

食べた (ta-beta) — “ate”

学ぶ (mana-bu) — “to study”

The べる, べた, and ぶ aren’t optional decoration — they’re what makes the verb conjugate, and their presence is a strong signal you’re looking at a kunyomi reading.

The rule of thumb (and how far it gets you)

Put together, here’s the pattern that predicts the large majority of real words:

This isn’t exceptionless — some compounds mix an onyomi and a kunyomi in the same word (called yutou or jubaku readings), and a handful of kanji break the pattern entirely. But as a first guess before reaching for a dictionary, this rule is right often enough to be genuinely useful, and it gets sharper the more real words you’ve read.

Why some kanji have several readings of each type

Common, high-frequency kanji tend to pick up multiple onyomi and multiple kunyomi over centuries of use — 生, for instance, has readings including sei, shou (onyomi) and i(kiru), u(mareru), nama (kunyomi), depending on the word. This looks intimidating listed all at once, but in practice you never learn them that way — you meet 生きる, 先生, 生まれる, and 生 (nama, “raw”) as separate words, in separate contexts, at separate times, and each reading anchors itself to the word it showed up in.

At the other extreme, some kanji have only one type:

• Onyomi-only kanji are usually specialised or scientific terms borrowed wholesale, with no equivalent native Japanese word ever existing.
• Kunyomi-only kanji include kokuji (国字) — characters actually invented in Japan, like 峠 (touge, “mountain pass”) — which were never part of the original Chinese set and so never picked up a Chinese-derived reading.

Why you shouldn’t memorise reading lists in isolation

It’s tempting to sit down with a chart of every reading for a kanji and try to memorise it cover to cover — resist that. Readings are far easier to retain when they’re anchored to an actual word you’ve read in context than when they’re abstract entries on a list. 食 means little as “shoku or ta or ku, take your pick”; 食べる and 食事 are two concrete, meaningful words that happen to share a character.

This is the same principle behind learning kanji inside words rather than as isolated shapes (covered in the kanji guide) — and it applies just as much to readings as it does to meanings.

Build reading-recognition through real sentences

The fastest way to make onyomi and kunyomi second nature is to meet kanji repeatedly in real, connected sentences — where the context, the okurigana, and the surrounding words all reinforce which reading applies, without you having to consciously work it out.

That’s exactly what Shinobi’s graded stories are built for, with furigana on every kanji so you can confirm a reading instantly rather than guess-and-check. Begin at JLPT N5 stories where reading patterns are simple and clear, then climb to N4 and N3 as your recognition builds. New to reading entirely? Start with pre-N5 stories, or browse the full library.

Where to go next

Onyomi and kunyomi are one piece of the kanji puzzle — see the full learning strategy in how to learn kanji, understand the role furigana plays in reading unfamiliar readings in our furigana guide, and once your foundation is solid, put it into practice by reading daily at your level.