Purpose The aim of this study was to evaluate the supportive role of molecular and structural biomarkers (CSF protein levels, FDG PET and MRI) in the early differential diagnosis of dementia in a large sample of patients with neurodegenerative dementia, and in determining the risk of disease progression in subjects with mild cognitive impairment (MCI). Methods We evaluated the supportive role of CSF Aβ 42 , t-Tau, p-Tau levels, conventional brain MRI and visual assessment of FDG PET SPM t-maps in the early diagnosis of dementia and the evaluation of MCI progression. Results Diagnosis based on molecular biomarkers showed the best fit with the final diagnosis at a long follow-up. FDG PET SPM t-maps had the highest diagnostic accuracy in Alzheimer’s disease and in the differential diagnosis of non-Alzheimer’s disease dementias. The p-tau/Aβ 42 ratio was the only CSF biomarker providing a significant classification rate for Alzheimer’s disease. An Alzheimer’s disease-positive metabolic pattern as shown by FDG PET SPM in MCI was the best predictor of conversion to Alzheimer’s disease. Conclusion In this clinical setting, FDG PET SPM t-maps and the p-tau/Aβ 42 ratio improved clinical diagnostic accuracy, supporting the importance of these biomarkers in the emerging diagnostic criteria for Alzheimer’s disease dementia. FDG PET using SPM t-maps had the highest predictive value by identifying hypometabolic patterns in different neurodegenerative dementias and normal brain metabolism in MCI, confirming its additional crucial exclusionary role.