The channel hypothesis of Alzheimer's disease (AD) proposes that the beta-amyloid (Abeta) peptides which accumulate in plaques in the brain actually damage and/or kill neurons by forming ion channels. Evidence from a number of laboratories has demonstrated that Abeta peptides can form ion channels in lipid bilayers, liposomes, neurons, oocyctes, and endothelial cells. These channels possess distinct physiologic characteristics that would be consistent with their toxic properties. Abeta channels are heterogeneous in size, selectivity, blockade, and gating. They are generally large, voltage-independent, and relatively poorly selective amongst physiologic ions, admitting calcium ion (Ca(2+)), Na(+), K(+), Cs(+), Li(+), and possibly Cl(-). They are reversibly blocked by zinc ion (Zn(2+)), and tromethamine (tris), and irreversibly by aluminum ion (Al(3+)). Congo red inhibits channel formation, but does not block inserted channels. Although much evidence implicates Abeta peptides in the neurotoxicity of AD, no other toxic mechanism has been demonstrated to be the underlying etiology of AD. Channel formation by several other amyloid peptides lends credence to the notion that this is a critical mechanism of cytotoxicity.