In Alzheimer's disease, amyloid beta (Abeta) peptide is deposited in neuritic plaques in the brain. The Abeta peptide 1-42 or the fragment 25-35 are neurotoxic. We here review our recent explorations of the mechanisms of Abeta toxicity in hippocampal cultures. Abeta had no effect on intracellular calcium in neurons but caused striking changes in nearby astrocytes. The [Ca(2+)](c) signals started approximately 5-15 min after Abeta application and consisted of sporadic [Ca(2+)](c) pulses. These were entirely dependent on extracellular Ca(2+), independent of ER Ca(2+) stores and resulted from Ca(2+) influx, probably through Abeta-induced membrane channels. The Ca(2+) signals were closely associated with transient, episodic acidification which may reflect displacement of protons from binding sites or Ca(2+)/2H(+) exchange. Abeta caused an increased rate of generation of reactive oxygen species (ROS), also seen in astrocytes and not in neurons. The increased ROS generation was blocked by inhibitors of the NADPH oxidase, strongly suggesting that this enzyme, normally associated with immune cells, is expressed in astrocytes. ROS generation was also Ca(2+)-dependent, suggesting that Abeta activation of the enzyme may be secondary to the increase in [Ca(2+)](c). Abeta caused delayed neuronal death despite the fact that all responses were seen only in astrocytes. Neurons could not be protected by glutamate receptor antagonists, but were rescued by inhibition of the NADPH oxidase, by antioxidants and by increasing glutathione. These data suggest that Abeta causes Ca(2+)-dependent oxidative stress by activating an astrocytic NADPH oxidase, and that neuronal death follows through a failure of antioxidant support.