Ataxin-3 (AT3), a protein that causes spinocerebellar ataxia type 3, has a C-terminus containing a polyglutamine stretch, the length of which can be expanded in its pathological variants. Here, we report on the role of Cu(2+), Mn(2+), Zn(2+) and Al(3+) in the induction of defective protein structures and subsequent aggregation/fibrillogenesis of three different non-pathological forms of AT3, i.e. murine (Q6), human non-expanded (Q26) and human moderately expanded (Q36). AT3 variants showed an intrinsic propensity to misfolding/aggregation; on the other hand, Zn(2+) and Al(3+) strongly stimulated the amplitude and kinetics of these conformational conversions. While both metal ions induced a time-dependent aggregation into amyloid-like fibrillar forms, only small oligomers and/or short protofibrillar species were detected for AT3s alone. The rate and extent of the metal-induced aggregation/fibrillogenesis processes increased with the size of the polyglutamine stretch. Mn(2+) and Cu(2+) had no effect on (Q6) or actually prevented (Q26 and Q36) the AT3 structural transitions. The observation that Zn(2+) and Al(3+) promote AT3 fibrillogenesis is consistent with similar results found for other amyloidogenic molecules, such as beta-amyloid and prion proteins. Plausibly, these metal ions are a major common factor/cofactor in the etiopathogenesis of neurodegenerative diseases. Studies of liposomes as membrane models showed dramatic changes in the structural properties of the lipid bilayer in the presence of AT3, which were enhanced after supplementing the protein with Zn(2+) and Al(3+). This suggests that cell membranes could be a potential primary target in the ataxin-3 pathogenesis and metals could be a biological factor capable of modulating their interaction with AT3.