We hypothesized that word-stem completion (WSC) priming and perceptual identification (PI) priming, two types of repetition priming, rely on different cognitive and neural mechanisms: WSC priming on a modification mechanism that influences lexical retrieval, and PI priming on plasticity in pre-lexical perceptual systems. We compared the priming performance of the amnesic patient H.M. with words that came into common usage after the onset of his amnesia, and thus were novel to him (post-1965 words), and with familiar (pre-1953) words. We also tested age- and education-matched normal control subjects (NCS) and a patient with anterograde amnesia of recent onset (P.N.). The modification hypothesis predicted that H.M. would fail to show WSC priming with post-1965 words because pre-existing lexical representations of the test stimuli would be necessary for priming to occur. H.M.'s WSC priming score in the post-1965 condition did not differ from 0, and was inferior to the performance of NCS and of P.N. In contrast, H.M. displayed normal WSC priming in the pre-1953 condition. H.M. also showed robust and equivalent levels of PI priming in both conditions. A final experiment demonstrated preserved post-1965 word PI priming in H.M. when his baseline performance was matched with his post-1965 WSC priming baseline score. Our results challenge models that assume that most kinds of verbal repetition priming rely on the same or similar perceptual mechanisms.