The neural networks associated with processing related pairs of words forming literal, novel, and conventional metaphorical expressions and unrelated pairs of words were studied in a group of 15 normal adults using fMRI. Subjects read the four types of linguistic expressions and decided which relation exists between the two words (metaphoric, literal, or unrelated). According to the Graded Salience Hypothesis (GSH, ), which predicts a selective RH involvement in the processing of novel, nonsalient meanings, it is primarily the degree of meaning salience of a linguistic expression rather than literality or nonliterality, which modulates the degree of left hemisphere (LH) and right hemisphere (RH) processing of metaphors. In the present study, novel metaphorical expressions represented the nonsalient interpretations, whereas conventional metaphors and literal expressions represented the salient interpretations. A direct comparison of the novel metaphors vs. the conventional metaphors revealed significantly stronger activity in right posterior superior temporal sulcus, right inferior frontal gyrus, and left middle frontal gyrus. These results support the GSH and suggest a special role for the RH in processing novel metaphors. Furthermore, the right PSTS may be selectively involved in verbal creativity.