Parenteral immunization of transgenic mouse models of Alzheimer disease (AD) with synthetic amyloid beta-peptide (Abeta) prevented or reduced Abeta deposits and attenuated their memory and learning deficits. A clinical trial of immunization with synthetic Abeta, however, was halted due to brain inflammation, presumably induced by a toxic Abeta, T-cell- and/or Fc-mediated immune response. Another issue relating to such immunizations is that some AD patients may not be able to raise an adequate immune response to Abeta vaccination due to immunological tolerance or age-associated decline. Because peripheral administration of antibodies against Abeta also induced clearance of amyloid plaques in the model mice, injection of humanized Abeta antibodies has been proposed as a possible therapy for AD. By screening a human single-chain antibody (scFv) library for Abeta immunoreactivity, we have isolated a scFv that specifically reacts with oligomeric Abeta as well as amyloid plaques in the brain. The scFv inhibited Abeta amyloid fibril formation and Abeta-mediated cytotoxicity in vitro. We have tested the efficacy of the human scFv in a mouse model of AD (Tg2576 mice). Relative to control mice, injections of the scFv into the brain of Tg2576 mice reduced Abeta deposits. Because scFvs lack the Fc portion of the immunoglobulin molecule, human scFvs against Abeta may be useful to treat AD patients without eliciting brain inflammation.