An fMRI study of interleaved melody recognition was conducted to examine the neural basis of the bottom-up and top-down mechanisms involved in auditory stream segregation. Hemodynamic activity generated by a mixed sequence was recorded in eight listeners who were asked to recognize a target melody interleaved with distractor tones when the target was presented either before or after the composite sequence. fMRI results suggest that similar cortical networks were involved in both conditions, including bilaterally the auditory cortices within the superior temporal gyrus as well as the thalamus and the inferior frontal gyrus. However, when listeners heard the melody before they had to extract it from the mixture, neural activation in the inferior frontal operculum was significantly enhanced bilaterally; no change in auditory cortical activity was detected.