Objective: We determined the rate of falls among cognitively normal, community-dwelling older adults, some of whom had presumptive preclinical Alzheimer disease (AD) as detected by in vivo imaging of fibrillar amyloid plaques using Pittsburgh compound B (PiB) and PET and/or by assays of CSF to identify Ab42, tau, and phosphorylated tau. Methods: We conducted a 12-month prospective cohort study to examine the cumulative incidence of falls. Participants were evaluated clinically and underwent PiB PET imaging and lumbar puncture. Falls were reported monthly using an individualized calendar journal returned by mail. A Cox proportional hazards model was used to test whether time to first fall was associated with each biomarker and the ratio of CSF tau/Ab42 and CSF phosphorylated tau/Ab42, after adjustment for common fall risk factors. Results: The sample (n 5 125) was predominately female (62.4%) and white (96%) with a mean age of 74.4 years. When controlled for ability to perform activities of daily living, higher levels of PiB retention (hazard ratio 5 2.95 [95% confidence interval 1.01-6.45], p 5 0.05) and of CSF biomarker ratios (p , 0.001) were associated with a faster time to first fall. Conclusions: Presumptive preclinical AD is a risk factor for falls in older adults. This study suggests that subtle noncognitive changes that predispose older adults to falls are associated with AD and may precede detectable cognitive changes.