Neuropsychiatric disturbances are extremely common in Alzheimer's disease (AD), and represent integral features of the illness, as well as appropriate targets for therapy. We are interested in designing trials aimed at preventing or delaying the emergence of psychopathology in AD. For symptomatic treatment of agitation, mood stabilizers, particularly sodium valproate, have proved to be beneficial in some patients. Since these effects take several weeks to emerge, we considered that they might be dependent on potentially neuroprotective actions of valproate, such as inhibition of apoptosis and slowing of neurofibrillary tangle formation. In this article we present the rationale for testing the neuroprotective potential of valproate experimentally in mouse models of tauopathy and in a clinical trial of patients with AD who lack psychopathology at baseline. Together, these studies will provide important tests of the hypothesis that valproate, either through inhibition of tau phosphorylation or some other mechanism, is a useful therapeutic agent to modify disease progression in AD.