It is a concern for auditory fMRI studies that acoustic noise generated by the scanner produces an auditory response that can confound stimulus-induced activation. To establish how to minimize this problem, the present study mapped the time-course of the auditory response to a burst of acoustic scanner noise by employing a single-event method. Recorded bursts of scanner noise were interspersed with clustered-volume acquisitions at a range of stimulus-to-imaging delays to map the response with a temporal resolution of 1 sec. There were strong responses (1.5% signal change) to scanner noise in primary and secondary auditory cortex. In both cortical areas, the mean response rose to a peak by 4-5 sec after stimulus onset and decayed after a further 5-8 sec. The time course indicates that noise contamination in auditory fMRI can be substantially reduced by using a 9-12-sec repetition time, thus maximizing the dynamic range available for displaying the response to acoustical stimuli of interest.