Individuals with early Alzheimer’s disease (AD) suffer from a selective and profound failure to form new memories. A novel molecular mechanism with implications for therapeutics and diagnostics is now emerging in which the specificity of AD for memory derives from disruption of plasticity at synapses targeted by neurologically active Aβ oligomers. We have named these oligomers “ADDLs” (for pathogenic Aβ-derived diffusible ligands). ADDLs constitute metastable alternatives to the disease-defining Aβ fibrils deposited in amyloid plaques. In AD brain, ADDLs accumulate primarily as Aβ 12-mers (∼54 kDa). The same size oligomers occur in tg-mouse AD models; in mice, these 12-mers appear concomitantly with memory failure, consistent with the ability of ADDLs to inhibit long-term potentiation (LTP) and block reversal of long-term depression (LTD). Mechanistically, ADDLs are gain-of-function ligands that bind with specificity to particular synapses, targeting synaptic spines. Binding leads to a rapid and ectopic expression of the memory-linked immediate early gene Arc. Such aberrant accumulation has been linked by others to memory dysfunction in tg-Arc mouse models. Consistent with the expected consequences of Arc overexpression, ADDLs promote loss of surface NMDA receptors and anomalous spine morphology, which are responses expected to contribute to plasticity failure and memory dysfunction. Importantly, the attack on synapses provides a putative mechanism that unifies AD memory dysfunction with major features of AD neuropathology. Recent findings show ADDL binding instigates synapse loss, AD-type tau hyperphosphorylation, and generation of reactive oxygen species (ROS). Binding sites for ADDLs are at or in the close vicinity of NMDA receptors. Antibodies against external domains of NMDA receptors reduce ADDL binding and inhibit ADDL-stimulated ROS formation. The ROS response also is inhibited by memantine, an open-channel blocker of NMDA receptors recently approved for AD therapeutics. The ability of memantine to contravene the impact of ADDLs offers a new mechanism to explain why an NMDA receptor antagonist should improve memory function in AD patients. Elimination of ADDLs by vaccines now under development could provide the first AD treatments that are truly disease-modifying. In addition to establishing a molecular mechanism of significant value for AD therapeutics and diagnostics, studies of ADDL interactions with synaptic pathways and control mechanisms ultimately may provide new insights into the extraordinary complexities of physiological synaptic information storage.