This paper examines four accent patterns displayed by athematic nouns in Proto-Indo- European. Each accent pattern is distinguished by either alternating stress or vowel quality between "weak" forms (nominative, accusative, vocative) and "strong" forms. I argue that surface stress is the result of the interplay of the lexical accent specifications of the morphemes that compose the stem. The strong endings are classified as dominant and are thus responsible for the accent/ablaut alternations. Optimality Theory is used to provide a synchronic phonological analysis of athematic noun accent. The weak forms are accounted for with a ranking of faithfulness and alignment constraints, including a positional faithfulness ranking in which faithfulness to roots is preferred over faithfulness to derivational affixes. The strong endings, which are dominant, trigger antifaithfulness constraints (Alderete 1999), and so a new type of antifaithfulness constraint is introduced that works within inflectional paradigms, based on the Optimal Paradigms model (McCarthy 2005).