Background: Predementia is a heuristic umbrella concept to classify older adults with cognitive impairment who do not suffer dementia. Many diagnostic entities have been proposed to address this concept, but most of them have not had widespread acceptance. Aims: To review clinical definitions, epidemiologic data (prevalence, incidence) and rate of conversion to dementia of the main predementia constructs, with special interest in the two most frequently used: mild cognitive impairment (MCI) and minor neurocognitive disorder (miNCD). Methods: We have selected in three databases (MEDLINE, Web of Science and Google scholar) the references from inception to 31 December 2019 of relevant reviews, population and community-based surveys, and clinical series with >500 participants and >3 years follow-up as the best source of evidence. Main results: The history of predementia constructs shows that MCI is the most referred entity. It is widely recognized as a clinical syndrome harbinger of dementia of several etiologies, mainly MCI due to Alzheimer's disease. The operational definition of MCI has shortcomings: vagueness of its requirement of "preserved independence in functional abilities" and others. The recent miNCD construct presents analogous difficulties. Current data indicate that it is a stricter predementia condition, with lower prevalence than MCI, less sensitivity to cognitive decline and, possibly, higher conversion rate to dementia. Conclusions: MCI is a widely employed research and clinical entity. Preliminary data indicate that the clinical use of miNCD instead of MCI requires more scientific evidence. Both approaches have common limitations that need to be addressed.