ABSTRACT Primes that mismatch the targets on a phoneme (/ka'di/-/ka'do/) should produce responses intermediate between repeated (/ka'do/-/ka'do/) and unrelated (/bebe/-/ka'do/) primes. But what about primes and targets that mismatch on accentual cues (/kado/-/ ka'do/)? A repetition priming effect was observed on the N400 component with repeated primes eliciting less negativity than unrelated primes. A partial repetition priming effect was observed when the primes and targets mismatched on a phoneme. Phonemic mismatch primes elicited less negativity than unrelated primes, and more negativity than repeated primes. In contrast, no difference was observed between the repeated and the accentual mismatch primes. Hence, contrary to a mismatch on phonemic cues, a mismatch on accentual cues has no consequence on the magnitude of the repetition priming effect. This finding reinforces the view of an abstract lexicon in which acoustic details irrelevant for identification are stripped away before making contact with lexical representations.