Although much maligned, the amyloid-beta (Abeta) protein has been shown to possess a number of trophic properties that emanate from the protein's ability to bind Cu, Fe and Zn. Abeta belongs to a group of proteins that capture redox metal ions (even under mildly acidotic conditions), thereby preventing them from participating in redox cycling with other ligands. The coordination of Cu appears to be crucial for Abeta's own antioxidant activity that has been demonstrated both in vitro as well as in the brain, cerebrospinal fluid and plasma. The chelation of Cu by Abeta would therefore be predicted to dampen oxidative stress in the mildly acidotic and oxidative environment that accompanies acute brain trauma and Alzheimer's disease (AD). Given that oxidative stress promotes Abeta generation, the formation of diffuse amyloid plaques is likely to be a compensatory response to remove reactive oxygen species. This review weighs up the evidence supporting both the trophic and toxic properties of Abeta, and while evidence for direct Abeta neurotoxicity in vivo is scarce, we postulate that the product of Abeta's antioxidant activity, hydrogen peroxide (H(2)O(2)), is likely to mediate toxicity as the levels of this oxidant rise with the accumulation of Abeta in the AD brain. We propose that metal ion chelators, antioxidants, antiinflammatories and amyloid-lowering drugs that target the reduction of H(2)O(2) and/or Abeta generation may be efficacious in decreasing neurotoxicity. However, given the antioxidant activity of Abeta, we suggest that the excessive removal of Abeta may prevent adequate chelation of metal ions and removal of O(2)(-z.ccirf;), leading to enhanced, rather than reduced, neuronal oxidative stress.