Background: Repetitive transcranial magnetic stimulation (rTMS) is proved to be effective in facilitating stroke recovery. However, its therapeutic mechanism remains unclear. The present study aimed to investigate changes in white matter fractional anisotropy (FA) after excitatory rTMS to better understand its role in motor rehabilitation.Materials and Methods: Acute stroke patients with unilateral subcortical infarction in the middle cerebral artery territory were recruited. The patients were randomly divided into an rTMS treatment group and a sham group. The treatment group received a 10-day 5 HZ rTMS applied over the ipsilesional primary motor area beginning at about 4 days after stroke onset. The sham group received sham rTMS. Diffusion tensor imaging (DTI) data were collected in every patient before and after the rTMS or sham rTMS. Voxel-based analysis was used to study the difference in FA between the two groups. The trial of this article has been registered on the ClinicalTrials.gov and the identifier is NCT03163758.Results: Before the rTMS, there is no significant difference in FA between the two groups. Differently, after the treatment, the rTMS group showed increased FA in the contralesional corticospinal tract, the pontine crossing tract, the middle cerebellar peduncle, the contralesional superior cerebellar peduncle, the contralesional medial lemniscus, and the ipsilesional inferior cerebellar peduncle. These fasciculi comprise the cortex-pontine-cerebellum-cortex loop. Increased FA was also found in the body of corpus callosum and the contralesional cingulum of the treatment group compared with the sham.Conclusion: The greater connectivity of contralesional cortico-cerebellar loop and the strengthening of interhemispheric connection may reflect contralesional compensation facilitated by the excitatory rTMS, which gives us a clue to understand the therapeutic mechanism of rTMS.