Reliable biomarkers for Alzheimer's disease (AD) are highly needed in the clinic. As a fluid surrounding the brain and reflecting the major neuropathological features characteristic to the AD brain, the cerebrospinal fluid (CSF) provides the natural source for AD biomarkers. The expected use of an ideal AD biomarker is for the following purposes: (1) diagnosis, (2) prediction, (3) monitoring of disease progression, and (4) drug discovery. Review of the literature revealed that CSF analysis, specifically amyloid-beta (Aβ42, total (T)-tau, and phosphorylated (P)-tau, are reliable markers for AD diagnosis, even at very early stages, particularly vs. healthy controls, while more limited evidence for distinguishing from other dementias. As for prediction, abnormal CSF markers are predictors of cognitive decline in healthy subjects, converting from MCI to development of AD, and of the rate of cognitive decline in mild AD. Regarding monitoring disease progression, the use of CSF biomarkers does not seem very promising since a comparison of the marker levels between baseline and following years of follow-up revealed a remarkable stability of biomarker levels in CSF. As for the use in drug discovery, it is estimated that using CSF markers for the selection of subjects for clinical trials may reduce robustly sample size and trial costs. Yet, since no effective drug is currently available, the contribution of CSF AD biomarkers in drug discovery cannot be currently fully assessed. Nevertheless, testing CSF for evidence of CNS inflammation may help safety monitoring in AD clinical trials. Factors affecting CSF biomarker levels that should be taken into account are assay variability as well as effects of age, gender, apoE and other genetic variations, education, and time of day. Much effort has been and is still being dedicated into developing and validating CSF AD biomarkers by many centers in the world. Identifying additional CSF components, reflecting not only the lesions characteristic to AD (plaques and tangles) but also more functional and structural brain parameters, may provide a wider profile of the changes taking place in AD brains, and be further used as reliable CSF biomarkers for AD monitoring.