The development of therapies for Alzheimer's disease (AD) presents numerous challenges for physicians, researchers, and the pharmaceutical industry, with many drug candidates showing promise at one stage of clinical research only to fall at the next hurdle. A great number of drugs with a variety of targets and clusters of mechanisms are currently in various stages of basic and clinical investigation. However, these hypothesis-derived agents may be tested much too late in the chronically progressive disease process to demonstrate meaningful effects or outcomes, mirroring the clinical syndromal scenario in which the underlying pathophysiological disease condition is frequently diagnosed extremely late. Moreover, the complexity of the disease calls for developments and improvements in study designs and methods modeled for different target populations and disease stages (e.g. asymptomatic to prodromal to syndromal). New integrated concepts and models of disease pathophysiology, use of validated and qualified biomarkers, outcomes and endpoints, particularly the development of a surrogate outcome, may allow targeting of characteristic mechanism-derived therapies of specifically affected biological systems at different time-points in the disease process, providing increasing opportunities for early and preventative intervention. A core set of feasible diagnostic and predictive biomarkers is already validated and in the process of standardization; however, continued and intensified research efforts will likely reveal a variety of novel biomarkers that grasp the complexity of the underlying disease process. In the future, trials of drugs to modify and prevent AD may embrace enrichment strategies and maybe be stratified by disease stage, genetic factors as well as by disease endophenotypes.