Objectives: Amyloid-β42 (Aβ42) appears central to Alzheimer's disease (AD) pathogenesis and is a major component of amyloid plaques. Mean cerebrospinal fluid (CSF) Aβ42 is decreased in dementia of the Alzheimer's type. This decrease may reflect plaques acting as an Aβ42 "sink," hindering transport of soluble Aβ42 between brain and CSF. We investigated this hypothesis. Methods: We compared the in vivo brain amyloid load (via positron emission tomography imaging of the amyloid-binding agent, Pittsburgh Compound-B [PIB]) with CSF Aβ42 and other measures (via enzyme-linked immunosorbent assay) in clinically characterized research subjects. Results: Subjects fell into two nonoverlapping groups: those with positive PIB binding had the lowest CSF Aβ42 level, and those with negative PIB binding had the highest CSF Aβ42 level. No relation was observed between PIB binding and CSF Aβ40, tau, phospho-tau 181, plasma Aβ40, or plasma Aβ42. Importantly, PIB binding and CSF Aβ42 did not consistently correspond with clinical diagnosis; three cognitively normal subjects were PIB-positive with low CSF Aβ42, suggesting the presence of amyloid in the absence of cognitive impairment (ie, preclinical AD). Interpretation: These observations suggest that brain amyloid deposition results in low CSF Aβ42, and that amyloid imaging and CSF Aβ42 may potentially serve as antecedent biomarkers of (preclinical) AD.