Transgenic Caenorhabditis elegans animals have been engineered to express wild-type and single-amino acid variants of a long form of human beta-amyloid peptide (A beta 1-42). These animals express high levels (approximately 300 ng of A beta/mg of total protein) of apparently full-length peptide, as determined by quantitative immunoblot. Expression of wild-type A beta in these animals leads to rapid production of amyloid deposits reactive with Congo red and thioflavin S. This model system has been used to examine the effect of Leu17Pro, Leu17Val, Ala30Pro, Met35Cys, and Met35Leu substitutions on the in vivo production of amyloid deposits. We find that the Leu17Pro and Met35Cys substitutions completely block the formation of thioflavin S-reactive deposits, implicating these as key residues for in vivo amyloid formation. We have also constructed transgenic strains expressing a novel A beta variant, the single-chain dimer. Animals expressing high levels of this variant also fail to produce thioflavin S-reactive deposits.