Oxidative stress, including the production of reactive oxygen species (ROS), has been reported to be a key event in the etiology of Alzheimer's disease (AD). Cu has been found in high concentrations in amyloid plaques, a hallmark of AD, where it is bound to the main constituent amyloid-beta (A beta) peptide. Whereas it has been proposed that Cu-A beta complexes catalyze the production of ROS via redox-cycling between the Cu(I) and Cu(II) state, the redox chemistry of Cu-A beta and the precise mechanism of redox reactions are still unclear. Because experiments indicate different coordination environments for Cu(II) and Cu(I), it is expected that the electron is not transferred between Cu-A beta and reactants in a straightforward manner but involves structural rearrangement. In this work the structures indicated by experimental data are modeled at the level of modern density-functional theory approximations. Possible pathways for Cu(II) reduction in different coordination sites are investigated by means of first-principles molecular dynamics simulations in the water solvent and at room temperature. The models of the ligand reorganization around Cu allow the proposal of a preferential mechanism for Cu-A beta complex reduction at physiological pH. Models reveal that for efficient reduction the deprotonated amide N in the Ala 2-Glu 3 peptide bond has to be protonated and that interactions in the second coordination sphere make important contributions to the reductive pathway, in particular the interaction between COO- and NH2 groups of Asp 1. The proposed mechanism is an important step forward to a clear understanding of the redox chemistry of Cu-A beta, a difficult task for spectroscopic approaches as the Cu-peptide interactions are weak and dynamical in nature.