If repetition priming is a perceptual phenomenon, why is it not elimin
ated when the modality is shifted between study and test? We investiga
ted cross-modal priming in a densely amnesic subject (K.C.) with bilat
eral hippocampal lesions using a fragment completion task. After makin
g semantic judgments on words in the visual and auditory modality, K.C
. was asked to complete visual word fragments with the first word that
came to mind. On three separate occasions, K.C. showed robust within-
modality priming but no reliable cross-modal priming, suggesting that
the latter relies on brain structures damaged in this patient. Our dat
a are compatible with the notion that cross-modal priming in healthy i
ndividuals reflects explicit contamination from episodic memory. Condi
tions that compromise episodic memory, such as amnesia, can abolish th
e effect.