The fractional anisotropy (FA) is calculated by using diffusion tensor imaging (DTI) with multiple motion probing gradients (MPG). While FA has become a widely used tool to detect moderate changes in water diffusion in brain tissue, the measured value is sensitive to scan parameters (e.g. MPG-direction, signal to noise ratio, etc.). Therefore, it is paramount to address the reproducibility of DTI measurements among multiple centers. The purpose of this study was to assess the inter-center variability of FA. We studied five healthy volunteers who underwent DTI brain scanning three times at three different centers (I-III), each with a 1.5 T scanner having a different MPG-schema. Then, we compared the FA and eigenvalue from the three centers measured in seven brain regions: splenium of corpus callosum (CCs), genu of corpus callosum (CCg), putamen, posterior limb of internal capsule, cerebral peduncle, optic radiation, and middle cerebellar peduncle. At the CCs and CCg, there was a statistical difference (p<0.05) between center Iand center IIfor the same MPG-directions. Furthermore, at CCs and CCg, there was a statistical difference (p<0.05) between center II and center III for different MPG-directions. Conversely, no statistical differences were found between center I and center III for the different MPG-directions for all regions. These results indicate that the FA value was affected by the MPG-schema as well as by the MPG-directions.