Disease-modifying treatments for Alzheimer disease (AD) have focused mainly on reducing levels of amyloid-β (Aβ) in the brain. Some compounds have achieved this goal, but none has produced clinically meaningful results. Several methodological issues relating to clinical trials of these agents might explain this failure; an additional consideration is that the amyloid cascade hypothesis--which places amyloid plaques at the heart of AD pathogenesis--does not fully integrate a large body of data relevant to the emergence of clinical AD. Importantly, amyloid deposition is not strongly correlated with cognition in multivariate analyses, unlike hyperphosphorylated tau, neurofibrillary tangles, and synaptic and neuronal loss, which are closely associated with memory deficits. Targeting tau pathology, therefore, might be more clinically effective than Aβ-directed therapies. Furthermore, numerous immunization studies in animal models indicate that reduction of intracellular levels of tau and phosphorylated tau is possible, and is associated with improved cognitive performance. Several tau-related vaccines are in advanced preclinical stages and will soon enter clinical trials. In this article, we present a critical analysis of the failure of Aβ-directed therapies, discuss limitations of the amyloid cascade hypothesis, and suggest the potential value of tau-targeted therapy for AD.