The aim of this work was to investigate the hypothesis that multimodal MRI is able to detect the progressive disruption of volume and microstructure of subcortical structures in patients with amnestic mild cognitive impairment (a-MCI) and mild Alzheimer's disease (AD) in comparison with healthy controls (CTRL). We combined volumetric and diffusion tensor imaging (DTI) techniques in a cross-sectional study including 30 a-MCI, 30 AD patients, and 30 age-matched CTRL. We employed a fully automated model-based segmentation algorithm on 3 Tesla MRI anatomical images and accurate coregistration of DTI to anatomical images to extract regional values of DTI parameters. Both the hippocampi significantly and progressively decreased in volume from CTRL through MCI to AD. Both the thalami showed a progressive and significant decrease in volume from CTRL to AD. Mean diffusivity (MD) values increased progressively across the three groups in the bilateral hippocampus, amygdala, and in the right caudate. No differences in fractional anisotropy (FA) values were found. Two distinct but overlapping patterns of progression of structural (i.e., atrophy) and microstructural (i.e., MD increase) damage were observed. Particularly, the pattern of atrophy was mirrored by the increasing value of the averaged MD, which provided a further indicator of subtle tissue disruption in the hippocampal structure in mild AD patients. Combining different MRI modalities can allow identifying sensitive indicators of the subtle pathogenic mechanisms that occur in subcortical areas of AD patients.