A novel method for spatially mapping anisotropy/orientation coherence of the eigenvector is presented. By using an eigenvector-based approach, an intervoxel diffusion coherence (IVDC) index was used to quantify the coherence of the principal diffusion directions within a voxel neighborhood. This method may allow reconstruction of a whole brain map to be used for diagnostic purposes. The IVDC index is calculated by a scatter matrix-based method in a voxel-wise manner. A simulation was performed using two fiber populations crossing at various separation angles. We demonstrate that the IVDC index was more sensitive than fractional anisotropy (FA) to changes in separation between the fibers under a noise-free condition. Diffusion-tensor images of six healthy volunteers were acquired on a 3.0T MR imager. The FA, coherence index, and IVDC were then calculated. The results showed that IVDC improved the contrast in several brain areas including thalamus, middle cerebral peduncle, and pons. We therefore conclude that the IVDC index provides reliable and complementary information on water diffusion in the brain. It may be useful in white matter tractography, especially to determine the termination point of a trajectory.