Unipolar Depression is one of the leading causes of disability and loss of quality of life. Research on its pathogenesis could provide information about influencing pathological und physiological factors and possibly new options in treating the disease. The amygdala, part of the limbic system, has been identified as a part of neuroanatomical networks of depression. Morphometric measurement of the amygdala in MRI revealed differences in volume, comparing major depressive disorder (MDD) patients and healthy controls. Yet, the results have been inconsistent so far. Different approaches in segmentation, quality of MRI but also hitherto unknown other influencing factors on amygdala volume may account for these discrepancies. For this study, a new segmentation protocol for use in high definition 3-Tesla-MRI was defined. Neuroanatomical boundaries and landmarks of earlier segmentation protocols were taken into account. Using all three dimensions to simplify separation from other structures, a highly reliable and valid instrument for measuring the amygdala could be provided. The new protocol was applied on MR images of MDD patients and healthy controls. A significant bilateral reduction of amygdala volume in patients could be shown. Evidence suggests that duration of the illness is linked to the degree of right amygdala volume reduction. MRI morphometry is only able to deliver an indirect copy of the real neuroanatomy, e.g. there is a loss of quality as a consequence of digitalizing in MRI and of inaccuracy in segmentation. In spite of imprecise absolute measurement, relative connections in clinical clusters can be shown. With future further improvement of MRI technology and accompanying adapted segmentation protocols, the inaccuracy of measurement will be reduced and the distortion between the amygdala in vivo and its segmented image will be diminished. Segmentation protocols must probably be adapted with increasing quality of the utilized MRI. Standardizing protocols and measuring instruments however is desirable to allow comparing the results of different studies and facilitate identification of pathological patterns.