Amyloid polymorphism presents a challenge to physical theories of amyloid formation and stability. The amyloidogenic protein serum amyloid A (SAA) exhibits complex and unexplained structural polymorphism in its N-terminal fragments: the N-terminal 11-residue peptide (SAA1-11) forms left-handed helical fibrils, while extension by one residue (SAA1-12) produces a rare right-handed amyloid. In this study, we use a combination of vibrational spectroscopy and ultramicroscopy to examine fibrils of these peptides and their terminally acetylated and amidated variants, in an effort to uncover the physical basis for this effect. Raman spectroscopy and atomic force microscopy provide evidence that SAA1-12 forms a beta-helical fibril architecture, while SAA1-11 forms more typical stacked beta-sheets. Importantly, N-terminal acetylation blocks fibril formation by SAA1-12 with no effect on SAA1-11, while C-terminal amidation has nearly the opposite effect. Together, these data suggest distinct electrostatic interactions at the N-and C-termini stabilize the two fibril structures; we propose model fibril structures in which C-terminal extension changes the favored intermolecular interaction between peptide monomers from an Argl-C-terminus charge pair to an N-terminus C-terminus charge pair. This model suggests a general mechanism for charge-mediated amyloid polymorphism and may inform strategies for design of peptide-based nanomaterials stabilized by engineered intermolecular contacts.