Protein aggregation related toxicity is implicated in a variety of neurodegenerative diseases including Alzheimer's, Huntington's, prion and Amyotrophic Lateral Sclerosis (ALS). The proteins or peptides known to aggregate in disease are unrelated in their amino acid sequence and native structure but form structurally similar aggregates - amyloids. Studies outlined in this dissertation were aimed at uncovering the underlying biophysical mechanisms of amyloid formation, and lay the groundwork to develop rational strategies to combat neurodegenerative diseases. More than 100 point mutations in the homodimeric metalloenzyme Cu, Zn superoxide (SOD1) dismutase are involved in a genetically inherited familial form of ALS (FALS). We have discovered a mechanism of in vitro SOD1 aggregation in which the native SOD1 dimer dissociates, metals are lost from the monomers and the resulting apo-monomers oligomerize in a rate-limiting step. Further, we have computationally estimated that a majority of FALS-associated point mutants in SOD1 (70 out of the 75 studied) decrease dimer stability and/or increase dimer dissociation propensity. Thus, we have proposed that the underlying biophysical basis of FALS-linked SOD1 aggregation is the mutation-induced increase in the propensity to form apo-monomers. To uncover the molecular determinants of SOD1 apo-monomer oligomerization, the rate-limiting step in aggregation, we have developed two complementary in silico approaches: (a) we have identified sequence fragments of SOD1 that have a high self-association propensity, and (b) we have performed molecular dynamics simulations of model SOD1 monomer and dimer folding and misfolding. In both cases, we have identified key residue-residue interactions in SOD1 responsible for maintaining fidelity to its native state. We have proposed that the disruption of one or more of these key interactions ("hot spots") is implicated in non-native oligomerization. To understand the effect of FALS mutations on the key interactions involved in maintaining native-state fidelity, we have studied the nanosecond dynamics of wild type SOD1 and 3 FALS-associated mutant apo-dimers and apo-monomers. We found that in wild type SOD1 the motions of the dimer interface are mechanically coupled to the motions of the structurally distal metal-coordinating loops of both monomeric subunits. We further found that the strain induced in the protein by dimer dissociation, point mutations, or by exposure to high temperature is transmitted to a specific hairpin in the protein, previously found to be implicated in maintaining fold fidelity. The altered dynamics of mutant SOD1 dimers and monomers provides structural insights into the flexibility required for oligomerization. Collectively, findings in this dissertation have enhanced our understanding of the complex mechanisms of protein aggregation. Mechanisms established and structural insights obtained herein may facilitate rational design of small molecules to prevent protein aggregation, hence provide a therapeutic intervention strategy in neurodegenerative diseases.