Several previous studies have compared the blood oxygen level-dependent (BOLD) hemodynamic response (HDR) in healthy elderly subjects to the HDR in young subjects. Some studies have found a relative decreased amplitude in the elderly in the visual cortex, whereas other studies have found the elderly HDR amplitude in the visual cortex to be nearly identical to that in young subjects. A possible explanation for the different findings is that the peak voxel HDR is similar between the groups, but that the HDR in the group-averaged region-of-interest (ROI) is "washed out" by the inclusion of less significant voxels (due to a smaller extent of activation in the elderly) or by the inclusion of negative-peaking voxels. We tested this hypothesis using event-related functional magnetic resonance imaging (fMRI ). While undergoing fMRI, subjects performed a simple visual and motor task, pressing with their index fingers in response to visual presentation of the word tap. Data from 18 subjects, 8 young and 10 elderly, were analyzed. For each subject, a visual and a motor ROI was selected by choosing the most significant positive voxels within the anatomically defined ROI. This individual subject approach excluded both low-significance and negative-peaking voxels. Similar peaks were found for the elderly and the young subjects in both motor and visual regions and a more sustained BOLD response was found for the elderly in both regions. Additionally, as predicted, a greater percentage of voxels with a negative HDR was found for the elderly in the visual region; this finding was also replicated in our reanalysis of an independent fMRI and aging study from the fMRI Data Center. Functional neuroimaging observations of negative HDRs in visual areas have been interpreted as the effect of unconstrained processing during rest. Our results suggest that the elderly may have more unconstrained visual processing during the rest condition in the scanner. The observation that the group differences in the BOLD response are sensitive to voxel selection (e.g., inclusion of low-significance and/or negative voxels) underscores the importance of ROI selection criteria in the interpretation of fMRI studies using elderly populations.