The accumulation of beta-amyloid (A beta) in neuritic plaques is thought to be causative for the progression of Alzheimer's disease (AD). Recently, both active immunization and passive administration of A beta antibodies dramatically attenuated amyloid plaque deposition, neuritic dystrophy, astrogliosis and behaviour deficits in transgenic animals. In addition, we and others have found that titres of naturally occurring anti-A beta antibodies in the CSF of AD patients are significantly lower than those in age-matched controls. Treatment with intravenous immunoglobulins (a preparation that contained anti-A beta antibodies) significantly lowered CSF levels of A beta in non-demented patients. In this study, anti-A beta antibodies were isolated from immunoglobulin preparations and these anti-A beta antibodies strongly block fibril formation or disrupt formation of fibrilar structures. Furthermore, these antibodies almost completely prevented neurotoxicity of A beta. In contrast, immunoglobulins depleted of anti-A beta antibodies had little effect on A beta fibril formation or protection of neuronal cells. This study supports the findings that human anti-A beta antibodies may interfere with the pathogenesis of AD by more than one mechanism, and administration of polyclonal human anti-A beta antibodies isolated from plasma is a potential therapeutic agent to prevent or slow down disease progression.