Synaptic and axonal loss are two major mechanisms underlying Alzheimer's disease (AD) pathogenesis, and biomarkers reflecting changes in these cellular processes are needed for early diagnosis and monitoring the progression of AD. Contactin-2 is a synaptic and axonal membrane protein that interacts with proteins involved in the pathology of AD such as amyloid precursor protein (APP) and beta-secretase 1 (BACE1). We hypothesized that AD might be characterized by changes in contactin-2 levels in the cerebrospinal fluid (CSF) and brain tissue. Therefore, we aimed to investigate the levels of contactin-2 in the CSF and evaluate its relationship with disease pathology. Contactin-2 was measured in CSF from two cohorts (selected from the Amsterdam Dementia Cohort), comprising samples from controls (cohort 1, n = 28; cohort 2, n = 20) and AD (cohort 1, n = 36; cohort 2, n = 70) using an analytically validated commercial enzyme-linked immunosorbent assay (ELISA). The relationship of contactin-2 with cognitive decline (Mini-Mental State Examination (MMSE)) and other CSF biomarkers reflecting AD pathology were analyzed. We further characterized the expression of contactin-2 in postmortem AD human brain (n = 14) versus nondemented controls (n = 9). CSF contactin-2 was approximately 1.3-fold reduced in AD patients compared with controls (p < 0.0001). Overall, contactin-2 levels correlated with MMSE scores (r = 0.35, p = 0.004). We observed that CSF contactin-2 correlated with the levels of phosphorylated tau within the control (r = 0.46, p < 0.05) and AD groups (r = 0.31, p < 0.05). Contactin-2 also correlated strongly with another synaptic biomarker, neurogranin (control: r = 0.62, p < 0.05; AD: r = 0.60, p < 0.01), and BACE1, a contactin-2 processing enzyme (control: r = 0.64, p < 0.01; AD: r = 0.46, p < 0.05). Results were further validated in a second cohort (p < 0.01). Immunohistochemical analysis revealed that contactin-2 is expressed in the extracellular matrix. Lower levels of contactin-2 were specifically found in and around amyloid plaques in AD hippocampus and temporal cortex. Taken together, these data reveal that the contactin-2 changes observed in tissues are reflected in CSF, suggesting that decreased contactin-2 CSF levels might be a biomarker reflecting synaptic or axonal loss.