Thirty-two studies (2889 subjects) that investigated the prevalence of the causes of dementia were critically reviewed. Particular attention was paid to potential and actual reversibility. Although dementia manifests itself primarily in old age (particularly age 75 and older), the mean age of patients for the studies that reported age data (56%) was 72.3 years. Twenty-five studies originated from secondary or tertiary centers, and four were community-based. Dementias consisted of Alzheimer disease, 56.8%; multi-infarct, 13.3%; depression, 4.5%; alcoholic, 4.2%; and drugs, 1.5%. No single other cause contributed more than 1.6% of the cases. Potentially reversible causes made up 13.2% of all cases. However, the more important question of whether patients with potentially reversible causes were followed and reversal actually seen was not always examined. In 11 studies (34%) that provided follow-up, 11% of dementias resolved, either partially (8%) or fully (3%). The commonest reversible causes were drugs, 28.2%; depression, 26.2%; and metabolic, 15.5%. Due to the presence of various biases (selection, lack of "blinded" investigators, and others) in the surveyed works, it is probable that the true incidence of reversible dementias in the community is even lower than that reported. Research implications as well as a conservative approach to the workup of a new case of dementia are offered.