Accumulating evidence supports the hypothesis that white matter (WM) abnormalities are involved in the pathophysiology of bulimia nervosa (BN); however, findings from in vivo neuroimaging studies have been inconsistent. We aimed to investigate the possible brain WM alterations, including WM volume and microstructure, in patients with BN. We recruited 43 BN patients and 31 healthy controls (HCs). All participants underwent structural and diffusion tensor imaging. Differences in WM volume and microstructure were evaluated using voxel‐based morphometry, tract‐based spatial statistics, and automated fibre quantification analysis. Compared with HCs, BN patients showed significantly decreased fractional anisotropy in the middle part of the corpus callosum (nodes 31–32) and increased mean diffusivity in the right cranial nerve V (CN V) (nodes 27–33 and nodes 55–88) and vertical occipital fasciculus (VOF) (nodes 58–85). Moreover, we found decreased axial diffusivity in the right inferior fronto‐occipital fasciculus (node 67) and increased radial diffusivity in the CN V (nodes 22–34 and nodes 52–89) and left VOF (nodes 60–66 and nodes 81–85). Meanwhile, WM microstructural changes were correlated with patients' clinical manifestations. We did not find any significant differences in WM volume and the main WM fibre bundle properties between BN patients and HCs. Taken together, these findings provide that BN shows significant brain WM reorganization, but primarily in microstructure (part of WM fibre bundle), which is not sufficient to cause changes in WM volume. The automated fibre quantification analysis could be more sensitive to detect the subtle pathological changes in a point or segment of the WM fibre bundle. We combined voxel‐based morphometry, tract‐based spatial statistics, and automated fibre quantification (AFQ) to analyse the whole brain white matter (WM) reorganization of BN patients. Bulimia nervosa (BN) could result in significantly abnormal WM alterations but mainly located in the WM microstructure, rather than WM volume and principal WM fibre bundles. The AFQ analysis may be more sensitive to detect the subtle pathological changes in a point or segment of the WM fibre bundle.