This paper presents an automatic fractional anisotropy (FA) asymmetry analysis and applies it to determine the FA asymmetry (FAA) changes associated with the sides of seizure origin of patients with temporal lobe epilepsy (TLE) using diffusion tensor imaging (DTI). All the control and patient images are first normalized onto the JHU-DTI-MNI atlas using a simultaneous deformable DTI registration algorithm, and the FA images are warped accordingly. Then, the tract-based spatial statistics (TBSS) algorithm is employed to quantify the FA on white matter (WM) skeletons, which are divided into different sections by overlapping with the 102 WM regions defined by the atlas. The FAA values, i.e., the relative differences of each WM skeleton section pair, are calculated. Statistical analysis is then performed to identify the regions that significantly contributed to the group differences between control and left/right TLE, as well as between left and right TLE. The results indicate that FAA correlates with the side of seizure origin, and those of certain regions are significantly different between normal controls and left or right TLE. The quantitative results can be useful for pre-surgical evaluation of TLE patients and for better understanding of the relationship between fiber tracts with the site of origin of TLE, EEG tests, the syndromes and neural psychological responses.