Available lysine losses from cottonseed, yeast, bacterial, algal, casein, and purified bacterial protein sources were monitored following thermal processing. A model system consisting of sufficient protein source to yield 15% protein, 4% glucose, and microcrystalline cellulose was used. The model systems were prepared by mixing the protein, glucose, and cellulose with phosphate buffer to obtain a slurry with a pH of 7.0, freeze-drying, and then rehydrating in a desiccator over saturated cupric chloride to a water activity of 0.68. The model systems were heated for 2 min at 130 Department of Nutrition (K.W.B.), Division of Environmental Studies (S.W.E., C.R.G.), and the Department of Animal Science (J.G.M.), University of California, Davis, California 95616. The Procter and Gamble Company, Miami Valley Laboratories, Cincinnati, Ohio 45247.C. Available lysine losses, as measured by Carpenter's fluorodinitrobenzene method, were dependent upon the protein source. Casein, which was included as a reference protein, and cottonseed exhibited the greatest losses of available lysine after processing followed by algal, yeast, and bacterial protein. In contrast, purified bacterial protein lost a greater proportion of its available lysine than any of the other protein sources.