Aggregation of amyloid-beta (Abeta) peptides is a central phenomenon in Alzheimer's disease. Zn(II) and Cu(II) have profound effects on Abeta aggregation; however, their impact on amyloidogenesis is unclear. Here we show that Zn(II) and Cu(II) inhibit Abeta(42) fibrillization and initiate formation of non-fibrillar Abeta(42) aggregates, and that the inhibitory effect of Zn(II) (IC(50) = 1.8 micromol/L) is three times stronger than that of Cu(II). Medium and high-affinity metal chelators including metallothioneins prevented metal-induced Abeta(42) aggregation. Moreover, their addition to preformed aggregates initiated fast Abeta(42) fibrillization. Upon prolonged incubation the metal-induced aggregates also transformed spontaneously into fibrils, that appear to represent the most stable state of Abeta(42). H13A and H14A mutations in Abeta(42) reduced the inhibitory effect of metal ions, whereas an H6A mutation had no significant impact. We suggest that metal binding by H13 and H14 prevents the formation of a cross-beta core structure within region 10-23 of the amyloid fibril. Cu(II)-Abeta(42) aggregates were neurotoxic to neurons in vitro only in the presence of ascorbate, whereas monomers and Zn(II)-Abeta(42) aggregates were non-toxic. Disturbed metal homeostasis in the vicinity of zinc-enriched neurons might pre-dispose formation of metal-induced Abeta aggregates, subsequent fibrillization of which can lead to amyloid formation. The molecular background underlying metal-chelating therapies for Alzheimer's disease is discussed in this light.