Most analyses of pre-nuclear glides in Chinese assume a division of the syllable into Onset and Rime, and focus on which of these the glide is in. I shall argue that there is little evidence to support such a constituent structure, and that in fact it leads to serious problems because for many speakers the glide does not behave consistently as a member of either constituent. A moraic model, with less structure, avoids most of these problems and accommodates the data more straightforwardly. The analysis depends crucially on a theory of markedness, and is couched in Optimality Theory. It takes a cross-linguistic perspective, and includes detailed analyses of Mandarin and English.