Multi-site study gives large statistical power for the results. Some intrinsic differences among scanners are recognizing as an inevitable property. Isotropic diffusion has already been discussing. In contrast, anisotropic diffusion has been evading the endeavor because of the absence of an easily available anisotropic diffusion phantom. We compared DTI measures among five different MR scanners at three different sites using commercially and easily available astriction cotton. The averaged coefficient of variation for longitudinal repeated acquisitions of DTI on five different MR scanners were stable. The scan-rescan properties of five different MR scanners can be comparable by anisotropic diffusion phantom.