Recent epidemiological, biological and genetic data indicate a relationship between cholesterol and Alzheimer's disease (AD) including the association of polymorphisms of ABCA1 (a gene that is known to participate in cholesterol and phospholipid transport) with AD prevalence. Based on these data, we postulated that genetic variation in the related and brain-specific ABCA2 gene leads to increase risk of AD. A large case-control study was conducted where the sample was randomly divided into a hypothesis-testing sample (230 cases/286 controls) and a validation sample (210 cases/233 controls). Among the 45 SNPs we tested, one synonymous SNP (rs908832) was found significantly associated with AD in both samples. Additional analyses performed on the whole sample showed a very strong association between this marker and early-onset AD (OR = 3.82, 95% C.I. = [2.00 - 7.30], P = 5 x 10(-5)). Further research is needed to understand the functional role of this polymorphism. However, together with the reported associations of AD with APOE, CYP46A1 and ABCA1, the present result adds a very significant support for the role of cholesterol and phospholipid homeostasis in AD and a rationale for testing novel cholesterol homeostasis-related therapeutic strategies in AD.