The amyloid β (Aβ) peptide, correlated with development of Alzheimer's disease (AD), is produced by sequential proteolytic cleavage of the amyloid precursor protein (APP) by β- and γ-secretases. Alternative proteolytic cleavage of APP by α-secretase prevents formation of Aβ peptide and produces a neuroprotective protein, a soluble fragment of APPα (sAPPα). We previously generated a single-chain variable domain antibody fragment (scFv) that binds APP at the β-secretase cleavage site and blocks cleavage of APP (iBsec1), and a second scFv which has been engineered to have α-secretase-like activity that increases α-secretase cleavage of APP (Asec1a) and showed that a bispecific antibody (Diab) combining both iBsec1 and Asec1a constructs protects mammalian cells from oxidative stress. Here, we show that the diabody is an effective therapeutic agent in a mouse model of AD. An apolipoprotein B (ApoB) binding domain peptide was genetically added to the diabody to facilitate transfer across the blood-brain barrier, and a recombinant human adeno-associated virus 2/8 (rAAV2/8) was used as a vector to express the gene constructs in a APP/PS1 mouse model of AD. The diabody increased levels of sAPPα, decreased Aβ deposits and levels of oligomeric Aβ, increased neuronal health as indicated by MAP2 and synaptophysin staining, increased hippocampal neurogenesis, and most importantly dramatically increased survival rates compared with untreated mice or mice treated only with the β-secretase inhibitor. These results indicate that altering APP processing to inhibit β-site activity while simultaneously promoting α-secretase processing provides substantially increased neuronal benefits compared with inhibition of β-secretase processing alone and represents a promising new therapeutic approach for treating AD.