This article presents an argument for the idea that consonants of high sonority are preferred over low-sonority consonants between vowels (Uffmann 2007). The argument is based on a detailed description and analysis of the phonological patterns which ensure that a syllable starts with a consonant in Washo. Washo inserts [j], which is of high sonority, between any two vowels. Word-initially however a glottal stop is inserted. It is argued that the Washo insertion patterns show influence of the sonority requirements, and these patterns are not subject to an analysis in terms of autosegmental spreading. Thus although feature spreading is a useful tool for capturing similarity requirements on neighboring segments, it is not sufficient to account for cases like Washo epenthesis.