In vitro assay procedures were applied in the measurement of available amino acids in a selection of fish meals representing good- and poor-quality product. Results were assessed by comparing them with results from chick-growth assays. Available methionine and tryptophan were assayed microbiologically with Streptococcus zymogenes, after predigestion of the test samples with papain or pronase. Results for methionine were correlated with chick-growth assays (r 0.86 for papain, 0.91 for pronase; P less than 0.01). Compared with the chick assays, papain tended to give lower, and pronase higher, results. Finer milling of the test samples did not influence the pronase values. Results for available tryptophan were also correlated with chick-growth assays (r 0.95 for papain, 0.96 for pronase; P less than 0.001). Compared with the chick values, papain gave markedly lower results and pronase marginally higher ones. Finer milling of the test samples increased the papain values by about 50% but had no effect with pronase. Available lysine was assayed microbiologically with Tetrahymena pyriformis and with a dye-binding procedure (DBL). The results correlated with the chick-growth assays (r 0.99 for DBL, P less than 0.001; 0.85 for Tetrahymena, P less than 0.01) but both methods overrated the poorer-quality samples. True nitrogen digestibilities and amino acid digestibilities were determined with chickens by the 'ileal analysis' procedure: the amino acid digestibilities were significantly higher and similar to the corresponding availabilities as measured in chick-growth assays. Ball milling a poor-quality fish meal caused a marked fall in its N digestibility, whereas similar treatment of a good-quality meal caused a slight increase. An explanation for this finding is proposed. Strep. zymogenes assays following pronase digestion of the test samples gave precise and acceptably accurate measures of the biologically available methionine and tryptophan in the test samples. For available lysine, Tetrahymena and DBL values for the poorer-quality samples were notably higher than the chick-growth assay; possible reasons for this are discussed. The ileal analysis procedure underestimated true N digestibility.