Despite major scientific advance in understanding the clinical and neuropathological characteristics of Alzheimer disease (AD), its therapeutic intervention has been a major disappointment over more than a decade. Although disease-modifying (DM) therapies have been a popular topic for at least the same duration of time in AD research community, it remains unclear about what they really mean and how clinical trials for testing the DM efficacy can be optimally designed. This talk will discuss some of the most fundamental issues in AD research and the statistical consequences in designing DM trials. I will first review some of the recent development in AD research including the clinical and neuropathological correlation, the amyloid hypotheses, the concept of preclinical AD, and the role of imaging and fluid biomarkers. I will then discuss statistical issues in designing clinical trials to test the DM efficacy of a therapy, including the time window for intervention, the efficacy endpoints, optimum design parameters, and statistical power.