PET imaging using [(18)F]fluorodeoxyglucose (FDG) and [(11)C]Pittsburgh compound B (PIB) have been proposed as biomarkers of Alzheimer disease (AD), as have CSF measures of the 42 amino acid beta-amyloid protein (Abeta(1-42)) and total and phosphorylated tau (t-tau and p-tau). Relationships between biomarkers and with disease severity are incompletely understood. Ten subjects with AD, 11 control subjects, and 34 subjects with mild cognitive impairment from the Alzheimer's Disease Neuroimaging Initiative underwent clinical evaluation; CSF measurement of Abeta(1-42), t-tau, and p-tau; and PIB-PET and FDG-PET scanning. Data were analyzed using continuous regression and dichotomous outcomes with subjects classified as "positive" or "negative" for AD based on cutoffs established in patients with AD and controls from other cohorts. Dichotomous categorization showed substantial agreement between PIB-PET and CSF Abeta(1-42) measures (91% agreement, kappa = 0.74), modest agreement between PIB-PET and p-tau (76% agreement, kappa = 0.50), and minimal agreement for other comparisons (kappa <0.3). Mini-Mental State Examination score was significantly correlated with FDG-PET but not with PIB-PET or CSF Abeta(1-42). Regression models adjusted for diagnosis showed that PIB-PET was significantly correlated with Abeta(1-42), t-tau, and p-tau(181p), whereas FDG-PET was correlated only with Abeta(1-42). PET and CSF biomarkers of Abeta agree with one another but are not related to cognitive impairment. [(18)F]fluorodeoxyglucose-PET is modestly related to other biomarkers but is better related to cognition. Different biomarkers for Alzheimer disease provide different information from one another that is likely to be complementary.