BACKGROUND AND OBJECTIVES: It can be clinically challenging to differentiate dementia with Lewy bodies (DLB) and Alzheimer disease (AD). As potential therapies emerge with the goal of slowing or halting misfolded protein aggregation, it is imperative to be able to identify individuals before the disease becomes disabling. Differentiating between DLB and AD in the preclinical or prodromal phase of DLB and AD becomes more important. Studies are needed to validate the proposed criteria for prodromal DLB. METHODS: Longitudinal data were obtained from the Uniform Data Set of the National Alzheimer's Coordinating Center. Included participants had a baseline diagnosis of normal or mild cognitive impairment and a consecutive 2-year follow-up diagnosis of DLB or AD. We examined whether core DLB clinical features, supportive neuropsychiatric features, and neuropsychological data in the 2 years preceding the dementia diagnosis distinguished DLB from AD. RESULTS: We identified 143 participants with DLB and 429 age-matched/sex-matched participants with AD. The presence of 2 or more core DLB features in the year before dementia diagnosis yielded the greatest AUC (0.793; 95% CI 0.748-0.839) in distinguishing prodromal DLB from prodromal AD. Sleep disturbances, hallucinations, and a cognitive profile of worse processing speed, attention, and visuoconstruction performance were evident at least 2 years before the dementia diagnosis in DLB compared with AD. DISCUSSION: Data from this multisite, longitudinal, well-characterized research North American cohort support the validity of the recently published criteria for prodromal DLB. In the prodromal stage, patients who subsequently develop DLB are more likely to have core DLB clinical features and worse attention, processing speed, and visuospatial performance than those who go on to develop AD. Differentiation of DLB and AD before dementia emerges provides an opportunity for early, disease-specific intervention and overall management.