To examine priming for novel, nonverbal patterns, a modification of the paradigm used by Gabrieli et al. (1990, Neuropsychologia, 28, 417–427) was administered to a group of Korsakoff patients, a group of alcoholic controls, and a group of normal controls. Subjects were asked to connect five-dot configurations into the first pattern that came to mind. Priming was than assessed by having the subjects copy experimenter-provided interpretations for each configuration and examining the effect of this manipulation on the subjects' subsequent interpretation of the same configurations. For the Korsakoff patients, the copied prime replaced their initial perceptual interpretation less frequently than it did for normal controls. Instead, the prime had its effect by combining with the baseline percept. These findings suggest that priming for novel, nonverbal material is weaker and less direct for Korsakoff patients than it is for controls.