The purpose of our study was to evaluate the spatial distribution of a gradient-pulse-to-acoustic-noise transfer function (GPAN-TF) with and without a phantom inside the bore. This study showed that the spatial distribution of GPAN-TFs with a phantom in the bore was significantly larger than that with an empty imager. The GPAN-TF spectrum for high frequency range were increased by the phantom. Therefore, the patient would be exposed to a more unpleasant sound than conventional evaluation in an empty scanner. The multipoint analysis using GPAN-TFs revealed structural differences in respective gradient coils under a situation similar to actual MR examination.