The purpose of this paper was to evaluate a phonological treatment program of maximal rather than minimal feature contrasts by charting the course of learning in a child displaying a systematic error pattern involving the nonoccurrence of word-initial consonants. Generalization data indicated that the child learned 16 word-initial consonants following treatment of only three sets of maximal opposition contrasts. Overgeneralization data indicated that the child restructured his phonological system based on a larger concept of "word initialness." Basic components of and differences between various forms of contrast treatment are discussed.