Solvational perturbations were employed to selectively tune the aggregational preferences of insulin at 60 degrees C in vitro in purely aqueous acidic solution and in the presence of the model co-solvent ethanol (EtOH) (at 40%(w/w)). Dynamic light scattering (DLS), thioflavin T (ThT)-fluorescence, Fourier transform infrared (FTIR) and atomic force microscopy (AFM) techniques were employed to characterize these pathways biophysically with respect to the pre-aggregational assembly of the protein, the aggregation kinetics, and finally the aggregate secondary structure and morphology. Using cell viability assays, the results were subsequently correlated with the cytotoxicity of the insulin species that form in the two distinct aggregation pathways. In the cosolvent-free solution, predominantly dimeric insulin self-assembles via the well-known amyloidogenic pathway, yielding exclusively fibrillar aggregates, whereas in the solution containing EtOH, the aggregation of predominantly monomeric insulin proceeds via a pathway that leads to exclusively non-fibrillar, amorphous aggregates. Initially present native insulin assemblies as well as partially unfolded monomeric species and low molecular mass oligomeric aggregates could be ruled out as direct and major cytotoxic species. Apart from the slower overall aggregation kinetics under amorphous aggregate promoting conditions, which is due to the chaotropic nature of high EtOH concentrations, however, both pathways were unexpectedly found to evoke insulin aggregates that were cytotoxic to cultured rat insulinoma cells. The observed kinetics of the decrease of cell viabilities correlated well with the results of the DLS, ThT, FTIR and AFM studies, revealing that the formation of cytotoxic species correlated well with the formation of large-sized, beta-sheet-rich assemblies (>500 nm) of both fibrillar and amorphous nature. These results suggest that large-sized, beta-sheet-rich insulin assemblies of both fibrillar and amorphous nature are toxic to pancreatic beta-cells. In the light of the ongoing discussion about putative cytotoxic effects of prefibrillar and fibrillar amyloid aggregates, our results support the hypothesis that, in the case of insulin, factors other than the specific secondary or quarternary structural features of the various different aggregates may define their cytotoxic properties. Two such factors might be the aggregate size and the aggregate propensity to expose hydrophobic surfaces to a polar environment.