Molecular imaging using nuclear medicine methods, i.e. positron emission tomography (PET) and single photon emission computed tomography (SPECT), is increasingly used in both clinical practice and in research for the investigation of dementias. Dementias are a heterogeneous group of diseases characterized by the loss of cognitive abilities and by behavioral and psychological symptoms, and have an increasing prevalence in our aging population. They represent a urgent issue in global health, given their strong impact not only on patients, but also on caregivers, families and societies. The scientific progresses on dementia biology and the ongoing clinical trials based on disease modifying therapies have heightened the urgency to develop sensitive and reliable markers to diagnose and monitor dementia. Indeed, PET and SPECT imaging are able to assess, non-invasively, neuronal loss and dysfunction, functional compensation strategies as well as the involvement of specific receptors and neurotransmission systems. Thus, they provide biomarkers for disease, particularly helpful as diagnostic tools in the early and preclinical phases of disease. The studies presented in this thesis demonstrate that nuclear medicine imaging of brain glucose metabolism and dopamine transporters, routinely performed in clinical practice, can be effectively combined with a gain in the classification of patients with suspected dementia. These studies also show that specific tracers, allowing the measurement of enzymatic pathways such as cholinesterase activity, are promising markers for the early and differential diagnosis of dementias. Nuclear medicine imaging can also provide important insights on the pathophysiological processes underlying the development and the clinical expression of dementia. An example is the study of the “reserve capacity”: the ability of the brain to compensate for ongoing neurodegenerative processes. The reserve is associated with a high pre-morbid intelligence, educational and occupational attainment and active, stimulating lifestyle. We demonstrated that the activation of brain reserve can be documented by molecular imaging methods already in the early and prodromal phase of Alzheimer's Disease. We also explored the impact of genetic risk factors, such as the Apolipoprotein E genotype, on the expression of reserve, and the association between reserve and the modulation of a specific neurotransmission system, namely the cholinergic system. Finally, new opportunities in this field come from the technological developments integrating nuclear medicine tomographs with other imaging modalities. Given that PET and magnetic resonance imaging (MRI) are the modalities of choice for most indications in neuroimaging, the recently developed hybrid PET/MRI has a tremendous potential for molecular neuroimaging, and preliminary results are described here. The current knowledge provides the bases for future research, exploring an increasing number of molecular biomarkers, such as PET tracers for amyloid plaques, hypoxia and cholinergic receptors, that may ultimately be useful to predict with high levels of accuracy the transition from healthy aging, mild cognitive impairment and clinically overt dementia.