Alzheimer's disease (AD) is the major cause of dementia in the elderly. The biochemical changes that precede AD may be present up to 20 years before the clinical manifestation of the disease. The translational development of AD biomarkers may be theoretically achieved via two different strategies: the first strategy can be defined as 'knowledge-based' (deductive method), while the second one is a hypothesis-generating 'unbiased' approach (inductive strategy). The 'knowledge-based' approach relies on a direct understanding of the neuropathological processes that underlie the development of AD. In contrast, the 'unbiased' approach involves the use of modern techniques including proteomics and bioinformatics that allow unbiased investigations of numerous putative markers that may be informative with regard to AD. Cerebrospinal fluid (CSF) dosage of neuropathological AD-associated proteins has already been incorporated into the neurochemical diagnosis of AD, attesting the relevance of translational research. In the last few years, biomarker discovery research has successfully utilized genomics and proteomics for the identification of several promising molecular markers for AD. In the present article, we discuss the present state of the art and the future challenges in the search of CSF biomarkers for AD. Copyright © 2011 S. Karger AG, Basel.