Five experiments are described which detail both the structure of lexical representations accessed in visual word recognition and the method of parsing words in order to access these representations. All experiments employ the lexical decision task. The first three experiments provide support for the view that the lexical representation of a word is a representation termed here the Basic Orthographic Syllabic Structure (or BOSS) of that word rather than any articulatorily defined structure. The principle for determining the BOSS is based on both orthographic and morphological factors. Experiments 1 and 2 demonstrate that words are disrupted more when a division is made in the word according to its pronunciation than when it is made according to its BOSS. It is shown in Experiment 3 that final silent E's are not a part of the lexical representations of words and this is predicted by the BOSS principle. The last two experiments provide evidence for a parsing procedure for retrieval that is a reiterative left-to-right technique whereby lexical searches are made for combinations of letters that begin with the initial letter of the word. The results, in all, support a system of lexical access that is not mediated by phonological encoding. © 1979 Academic Press, Inc.