There has been evidence for a causal relationship between homocysteine and Alzheimer's disease for several years but the mechanism is unclear. In vivo, some homocysteine is converted to the thiolactone. This report describes a novel reaction between homocysteine thiolactone and dehydroascorbic acid in which the homocysteine thiolactone is converted to 3-mercaptopropionaldehyde. This product is shown to react with proteins causing their precipitation (probably by cross-linking). The two reactions are extremely facile and appear to be physiologically compatible suggesting a mechanism by which homocysteine may promote the deposition of proteins in nerve cells as amyloid plaques and fibrillary tangles.