Peripheral markers for Alzheimer's disease are of interest to confirm the diagnosis, to perform epidemiological screening, to identify distinct groups of patients, to predict the outcome of the disease, to monitor its progression and its sensibility to treatment and to give help in performing studies on the relationship between brain and behaviour and on the pathophysiology of the Alzheimer's disease. The ideal biomarker for Alzheimer's disease should detect a fundamental feature of neuropathology and be validated in neuropathologically confirmed cases and be confirmed by at least two independent studies; should be as sensitive and specific than the clinical diagnosis (about 85% and 80%), reliable, reproducible, simple to perform, inexpensive and non invasive (studies on blood, urine, saliva, or buccal scrapings) or moderately invasive (skin, rectal biopsies, bone marrow samples, or cerebrospinal fluid). Such a marker has not yet been found. In this paper we present those markers which come closest to fulfilling criteria for a useful biomarker, keeping in mind that these criteria depends on what purpose it is used (screening, prediction, diagnosis, monitoring, pathophysiological studies.) and that the finding of a good marker depends on the understanding of the disease.