Seventeen test foods were each analyzed by four methods. Total lysine was measured by conventional amino acid analysis. Reactive lysine was measured with either fluorodinitrobenzene, o-phthalaldehyde or a differential dye-binding procedure. The results were then compared with another group's results from rat growth assays of the same samples for availably lysine. A sample of deliberately heat-damaged milk powder gave a rat assay value corresponding to 64% of its total lysine content; other values were all higher and on average 99% for 7 animal products, and 87% for 9 vegetable products. The correlation coefficient between the two sets of values was 0.95. The 'reactive lysine' procedures failed to give a better prediction of the rat values.