New concepts about Alzheimer's disease (AD), considered as a clinical-biological entity, make essential the definition of biomarkers that could be used for the in vivo diagnosis of the disorder before dementia develops. Different types of genetic, biochemical and neuroimaging markers have been described, highlighting some of the changes that occur in the brain during the course of the disease, yet there is little proof of their pathognomonic and diagnostic value. Furthermore, many of the assays used are difficult to perform, the equipment/reagents are expensive or potentially hazardous (e.g.; use of radioactive compounds, CSF extraction). Thus, there is a need to define more suitable and convenient approaches, such as the determination of blood parameters that are easy to obtain and that can be repeated as necessary without contraindications. These data can be used by algorithms that combine specific and non-specific changes to classify patients at different stages of AD and/or distinguish AD from other related diseases with a greater specificity and reliability (> 80%). The blood parameters considered in this review are varied, including: β-amyloid, tau, apolipoproteins and proteins, as well as the metabolic behavior of blood cells, etc. Among the proteins, cytokines/chemokines and other cell factors related to both neuro-inflammatory and peripheral-inflammatory processes in AD are of prime importance. New technologies to detect and quantify these substances, reasonably priced such as the vibrational spectroscopy, panels of parameters and algorithms to assess the results, would be fundamental for the early AD diagnosis and to define new potential therapies.