Nine young right-handed men viewed colored pictures of people, scenes, and landscapes. Then, 24 hr later while undergoing PET scanning, they viewed previously studied (OLD) pictures in one type of scan, and previously not seen (NEW) pictures in another. The OLD-NEW subtraction of PET images indicates familiarity, and the NEW-OLD indicates novelty. Familiarity activations, signalling aspects of retrieval, were observed in the left and right frontal areas, and posterior regions bilaterally. Novelty activations were in the right limbic regions, and bilaterally in temporal and parietal regions, including area 37. These latter activations were located similarly to novelty activations in previous PET studies using visual words and auditory sentences, suggesting the existence of brain regions specializing in transmodal novelty assessment The effects of novelty are seen both be haviorally and in replicable patterns of cortical and subcortical activation. We propose a “novelty/encoding hypothesis”: (1) novelty assessment represents an early stage of long-term memory encoding; (2) elaborate, meaning-based encoding processes operate on the in coming information to the extent of its novelty, and therefore (3) the probability of long-term storage of information vanes directly with the novelty of the information.