This paper considers the phonological representation of nasals in Japanese; unlike the majority of other consonant classes, the syllabification of nasals varies in the literature. Most phonologists agree that the place-specified nasals are syllabified into onset positions, certain conditions excepted (Abe 1986, Itô 1986, Yoshida 1990), but the syllabification of the placeless nasal gives rise to some controversy. The latter is syllabified into a coda by Abe and Itô, but into a nucleus by Yoshida. In order to provide a non-arbitrary account of phonological processing with regard to nasals, I propose that the distribution of all nasals in Japanese is restricted to onsets exclusively; this is formulated in line with the Principle of Structure Preservation and the notion of phonetic interpretation.