The article discusses the potential issues with interpreting "axial" and "radial" diffusivities derived from the eigenvalues of the diffusion tensor in diffusion tensor imaging (DTI). These diffusivities are often used to infer biophysical properties like myelin and axonal density. However, the study shows that changes in "radial" diffusivity can lead to fictitious changes in "axial" diffusivity and vice versa, especially in voxels with crossing fibers. In vivo data from healthy and multiple sclerosis (MS) patients show that the angle between principal eigenvectors of corresponding voxels can exceed 45° in areas of low anisotropy, severe pathology, and partial volume. In some white matter pathology areas, "radial" diffusivity is 10% higher than in normal tissue, with principal eigenvector direction altered by more than 45°. This suggests that interpreting changes in "axial" and "radial" diffusivities based on tissue structure may be misleading without thorough investigation of their mathematical and geometrical properties. The study also highlights that the direction of the principal eigenvector is not always preserved in pathological tissue and may not align with the expected tissue architecture. Numerical simulations and in vivo data demonstrate that changes in microstructure can lead to unpredictable changes in measured diffusivities, unrelated to the underlying tissue organization. The study recommends caution in using "axial" and "radial" diffusivity terminology and instead referring to the eigenvalues of the DT. The findings emphasize the limitations of the tensor model and discourage the association between "radial" diffusivity and demyelination in complex tissue architecture. The study also presents a method to analyze in vivo data using corresponding voxels from different datasets to highlight areas where the direction of the principal eigenvector lies in different planes, potentially jeopardizing direct comparisons of "axial" and "radial" diffusivities. The results show that in MS patients, there are more voxels with significant changes in principal eigenvector direction and diffusivity compared to healthy controls. The study concludes that the eigenvalues of the DT do not necessarily reflect the same underlying structural characteristics in different datasets due to differences in the orientation of the principal eigenvector. This has implications for interpreting DTI-derived parameters in pathological conditions.The article discusses the potential issues with interpreting "axial" and "radial" diffusivities derived from the eigenvalues of the diffusion tensor in diffusion tensor imaging (DTI). These diffusivities are often used to infer biophysical properties like myelin and axonal density. However, the study shows that changes in "radial" diffusivity can lead to fictitious changes in "axial" diffusivity and vice versa, especially in voxels with crossing fibers. In vivo data from healthy and multiple sclerosis (MS) patients show that the angle between principal eigenvectors of corresponding voxels can exceed 45° in areas of low anisotropy, severe pathology, and partial volume. In some white matter pathology areas, "radial" diffusivity is 10% higher than in normal tissue, with principal eigenvector direction altered by more than 45°. This suggests that interpreting changes in "axial" and "radial" diffusivities based on tissue structure may be misleading without thorough investigation of their mathematical and geometrical properties. The study also highlights that the direction of the principal eigenvector is not always preserved in pathological tissue and may not align with the expected tissue architecture. Numerical simulations and in vivo data demonstrate that changes in microstructure can lead to unpredictable changes in measured diffusivities, unrelated to the underlying tissue organization. The study recommends caution in using "axial" and "radial" diffusivity terminology and instead referring to the eigenvalues of the DT. The findings emphasize the limitations of the tensor model and discourage the association between "radial" diffusivity and demyelination in complex tissue architecture. The study also presents a method to analyze in vivo data using corresponding voxels from different datasets to highlight areas where the direction of the principal eigenvector lies in different planes, potentially jeopardizing direct comparisons of "axial" and "radial" diffusivities. The results show that in MS patients, there are more voxels with significant changes in principal eigenvector direction and diffusivity compared to healthy controls. The study concludes that the eigenvalues of the DT do not necessarily reflect the same underlying structural characteristics in different datasets due to differences in the orientation of the principal eigenvector. This has implications for interpreting DTI-derived parameters in pathological conditions.