Certain phonological error patterns have been judged to be "unusual" or "idiosyncratic", posing a number of theoretical and clinical problems. This paper reconsiders an especially challenging case of an unusual error pattern documented by Leonard and Brown (1984). T (age 3;8) replaced all word-final consonants (except for labial stops) with [s] but more importantly inserted [s] after word-final vowels. Our purpose is to show that optimality theory offers a fresh perspective on this error pattern and its course of development by providing an account of the facts that is entirely consistent with general principles relevant to the description of fully developed languages and acquisition. (Contains 1 figure.)