In 2011, the so-called Dubois criteria introduced the use of biomarkers in research (in particular, brain amyloid positron emission tomography imaging and the cerebrospinal fluid levels of tau/fosfo-tau and beta-amyloid 1-42) for the early or preclinical diagnosis of Alzheimer's disease. Even so, we are looking at an increased use of these markers in clinical practice. In the 1960s, Alzheimer's disease was considered a rare form of presenile dementia, but gradually it has been recognized as the prevalent form of old-age dementia. As a consequence, what was once regarded as an inevitable outcome of old age is now recognized as a true disease. Several factors contributed to this paradigm shift, in particular a longer lifespan, new techniques of in vivo study of the central nervous system, and the pressure exerted by the pharmaceutical industry and patient groups. The current lack of disease-modifying therapies and the high incidence of mild cognitive impairment, which is a risk factor for dementia, raise a series of clinical ethical problems ranging from how diagnosis is communicated to how resources are used. This article offers a conceptual scheme through which these issues can be addressed.