Many previous studies suggested that both short-term and long-term motor training can modulate brain structures. However, little evidence exists for such brain anatomical changes in top-level gymnasts. Using diffusion-weighted and structural magnetic resonance images of the human brain, we applied voxel-based morphometry (VBM) and tract-based spatial statistics (TBSS) as well as FA-VBA (voxel-based analysis of fractional anisotropy, a VBM-style analysis) methods to quantitatively compare the brain structural differences between the world class gymnasts (WCG) and the non-athlete groups. In order to reduce the rate of false positive findings, we first determined that the clusters defined at a threshold of t > 2.3 and a cluster significance of p < 0.05 (FWE-corrected) across all subjects were the brain regions that showed significant differences in a between-group comparison. We then constructed several between-group comparisons based on the repeated diffusion or structural MRI data and created the intersecting regions from multiple between-group comparisons. Thus, we found significantly decreased fractional anisotropy (FA) not only in the white matter of the WCG in areas that included the bilateral superior longitudinal fasciculus, inferior longitudinal fasciculus, and inferior occipito-frontal fascicle, but also in the gray matter of the WCG in areas that included the bilateral middle cingulum, bilateral postcentral gyri, and bilateral motor regions. We also identified significantly increased gray matter density in the WCG, especially in the left inferior frontal gyrus, bilateral inferior and superior parietal lobule, bilateral superior lateral occipital cortex, left cuneus, left angular gyrus, and right postcentral gyrus. We speculate that the brain changes of the WCG may reflect the gymnasts' extraordinary ability to estimate the direction of their movements, their speed of execution, and their identification of their own and surrounding objects' locations. Our findings suggest that our method of constructing intersecting regions from multiple between-group comparison can considerably reduce the false positives, and our results provide new insights into the brain structure changes induced by long-term intensive gymnastic training.