This book presents evidence for a model of phonology in which words have both auditory and articulatory representations, with different constraints applying to each type of representation. The main constraints on auditory representations require contrasting sounds to be auditorily distinct from each other, i.e. these constraints implement a preference that contrasts should be easy for listeners to discriminate. This traditional notion is formalized in Optimality Theoretic terms as part of a general theory of the selection of contrasting sounds. The distinctiveness constraints interact with others (such as effort minimization and constraints relating to stress) to derive a variety of phonological phenomena, e.g. allophonic variation to maintain the distinctiveness of a contrast in different contexts, and neutralization in contexts where a contrast would be indistinct. The model is exemplified through the analysis of such patterns, drawn from a wide variety of languages. Outstanding Dissertations in Linguistics series.