Background and purpose: Biomarkers support the aetiological diagnosis of neurocognitive disorders in vivo. Incomplete evidence is available to drive clinical decisions; available diagnostic algorithms are generic and not very helpful in clinical practice. The aim was to develop a biomarker-based diagnostic algorithm for mild cognitive impairment patients, leveraging on knowledge from recognized national experts. Methods: With a Delphi procedure, experienced clinicians making variable use of biomarkers in clinical practice and representing five Italian scientific societies (neurology – Società Italiana di Neurologia per le Demenze; neuroradiology – Associazione Italiana di Neuroradiologia; biochemistry – Società Italiana di Biochimica Clinica; psychogeriatrics – Associazione Italiana di Psicogeriatria; nuclear medicine – Associazione Italiana di Medicina Nucleare) defined the theoretical framework, relevant literature, the diagnostic issues to be addressed and the diagnostic algorithm. An N–1 majority defined consensus achievement. Results: The panellists chose the 2011 National Institute on Aging and Alzheimer's Association diagnostic criteria as the reference theoretical framework and defined the algorithm in seven Delphi rounds. The algorithm includes baseline clinical and cognitive assessment, blood examination, and magnetic resonance imaging with exclusionary and inclusionary roles; dopamine transporter single-photon emission computed tomography (if no/unclear parkinsonism) or metaiodobenzylguanidine cardiac scintigraphy for suspected dementia with Lewy bodies with clear parkinsonism (round VII, votes (yes-no-abstained): 3-1-1); 18F-fluorodeoxyglucose positron emission tomography for suspected frontotemporal lobar degeneration and low diagnostic confidence of Alzheimer’s disease (round VII, 4-0-1); cerebrospinal fluid for suspected Alzheimer’s disease (round IV, 4-1-0); and amyloid positron emission tomography if cerebrospinal fluid was not possible/accepted (round V, 4-1-0) or inconclusive (round VI, 5-0-0). Conclusions: These consensus recommendations can guide clinicians in the biomarker-based aetiological diagnosis of mild cognitive impairment, whilst guidelines cannot be defined with evidence-to-decision procedures due to incomplete evidence.