Functional MRI is a popular tool for investigating central processing of visceral pain in healthy and clinical populations. Despite this, the reproducibility of the neural correlates of visceral sensation by use of functional MRI remains unclear. The aim of the present study was to address this issue. Seven healthy right-handed volunteers participated in the study. Blood oxygen level-dependent contrast images were acquired at 1.5 T while subjects received nonpainful and painful phasic balloon distensions ("on-off" block design, 10 stimuli per "on" period, 0.3 Hz) to the distal esophagus. This procedure was repeated on two further occasions to investigate reproducibility. Painful stimulation resulted in highly reproducible activation over three scanning sessions in the anterior insula, primary somatosensory cortex, and anterior cingulate cortex. A significant decrease in strength of activation occurred from session 1 to session 3 in the anterior cingulate cortex, primary somatosensory cortex, and supplementary motor cortex, which may be explained by an analogous decrease in pain ratings. Nonpainful stimulation activated similar brain regions to painful stimulation, but with greater variability in signal strength and regions of activation between scans. Painful stimulation of the esophagus produces robust activation in many brain regions. A decrease in subjective perception of pain and brain activity from the first to the final scan suggests that serial brain imaging studies may be affected by habituation. These findings indicate that for brain imaging studies that require serial scanning, development of experimental paradigms that control for the effect of habituation is necessary.