Romaji is the representation of Japanese sounds using Latin alphabet characters. Its origins are traced to the 16th century, when Portuguese missionaries devised a way to record the Japanese language. Today, romaji is widely used in international contexts such as passports, railway station signs, and road signs, providing a way to convey Japanese information to those unfamiliar with the language.

The two main romanization systems are Hepburn and Kunrei-shiki. Hepburn romanization is designed to be intuitive for English speakers and is used on passports and railway signage. For example, "shi" for し and "tsu" for つ. Kunrei-shiki, on the other hand, follows the phonological structure of Japanese more closely, writing "si" for し and "tu" for つ. Usage has been divided by context: Kunrei-shiki in Japanese-language textbooks, Hepburn in English textbooks and for proper nouns such as personal names and signage. The 1954 Cabinet notification that defined Kunrei-shiki was revised in 2025 to make Hepburn-style spellings the basis, so as of 2026 the two systems coexist in a transition period.

Japanese input methods (IME) primarily use romaji input, where typing Latin characters on a keyboard converts them to hiragana, which can then be further converted to kanji or katakana. Romaji input requires learning only the 26-key alphabet layout, making it easier to master than kana input (which uses the 50-sound layout). Kana input needs fewer keystrokes per character and can be faster once learned, but the burden of memorizing its layout keeps its users a minority, and romaji input has become the de facto standard for Japanese input.

Romaji is commonly used in URLs and filenames where Japanese characters would be impractical. Including Japanese directly in a URL results in lengthy percent-encoded strings, so creating slugs in romaji is standard practice. For instance, "Tokyo Tower" (東京タワー) becomes tokyo-tower in a URL, keeping it short and readable.

The differences between Hepburn and Kunrei-shiki can cause practical confusion. Passport names follow Hepburn by default, but discrepancies may arise with credit card or bank account names. Subtle rules such as long vowel notation ("o" vs. "ou" for おう) and the treatment of the nasal "n" before b/m/p ("m" vs. "n") also affect the resulting character count.

From a character counting perspective, romaji characters are counted as single half-width characters. The same Japanese text tends to have a higher character count in romaji than in hiragana. For example, "さくら" (sakura) is 3 characters in hiragana but 6 in romaji. In terms of bytes, hiragana uses 3 bytes per character in UTF-8 (9 bytes total), while romaji uses 1 byte per character (6 bytes total), making romaji more byte-efficient.