Ammonium phosphate labelled with $^{15}$N has been used in a single quantitative experiment to trace the pathways of ammonia assimilation and amino acid synthesis in food yeast. Methods have been developed and are briefly described, whereby the free amino acids and amides, and the amino acid residues of the proteins, may be extracted from the yeast, separated by ion-exchange chromatography, quantitatively estimated, and the nitrogen of their $\alpha $-amino groups specifically liberated for isotopic analysis.The yeast was cultured in shake-flasks on a minimal medium containing glucose, ammonium phosphate and mineral salts. By analysis of cells removed from the culture at various times it was shown that the percentage composition remained sensibly constant throughout the part of the exponential phase investigated, and hence the yeast was assumed to be in steady-state growth. For the isotopic experiment the yeast culture was transferred to medium containing ($^{15}$NH$_{4}$)$_{2}$HPO$_{4}$ and samples where then removed at intervals for analysis of the free and protein amino acids and for measurement of their $^{15}$N-abundance. After 30 min the remaining yeast was transferred back into unlabelled medium and further samples were then taken. This double-transfer procedure was used in order to permit more stringent tests of metabolic relationships to be made in the subsequent kinetic analysis.The quantitative analysis of the isotopic data was made by comparison with a model reaction system. The model consists of a series of branching reaction chains linking steady-state pools of intermediates from which material is randomly withdrawn in subsequent reactions; primary products of nitrogen assimilation can give rise to secondary and tertiary derivatives which, as amino acids, can act as precursors in protein synthesis. A series of kinetic equations have been derived, relating the isotopic abundance of a component in the model to the rates of the various reactions involved in its biosynthesis. By substituting numerical values in these equations and comparing the results with the experimental data it has been possible to assign to each amino acid a position in the model and to make an estimate of its rate of synthesis. This estimate can then, as a further test, be compared with the rate known to be necessary to maintain steady-state growth.The kinetic analysis indicates that glutamic acid and glutamine are the only amino acids to derive their $\alpha $-nitrogen directly from ammonia; they are synthesized at a rate sufficient to provide all the $\alpha $-amino nitrogen required for growth of the yeast but not to meet the total nitrogen requirements, so that other pathways for the assimilation of nitrogen must also operate. All the other amino acids apparently derive their $\alpha $-amino-N from glutamic acid, many of them directly. For some of the amino acids, the labelling of the residues in the protein is consistent with their having come directly from the pool of free amino acid; this emphasizes the very small size of any pools of intermediates between amino acid and protein. For other amino acids a more complex relationship has been observed between the free amino acid pool and the proteins; the data are best interpreted by assuming that not all of the pool is available as an intermediate in protein synthesis, some of it being spatially separated and not further metabolized. This separate pool is here called a storage pool and is envisaged as functioning as part of the regulatory mechanisms of the cell by removing any small overproduction of amino acid.The results are further considered in relation to known pathways of amino acid biosynthesis in micro-organisms. The data for alanine, aspartic acid, glycine, leucine, isoleucine, valine, tyrosine and phenylalanine are consistent with these amino acids, having been formed directly by transamination from glutamic acid. Similar transaminations, but followed by other reactions, can account for the synthesis of histidine, lysine, serine and methionine; there is no evidence for alanine-hydroxypyruvate transamination in serine synthesis or for the operation of the cystathionine pathway to methionine. Threonine does not apparently derive its nitrogen from aspartic acid in this experiment, and the operation of the pathway from aspartic acid via homoserine to threonine is questioned for yeast grown on a minimal medium. The very low isotopic abundance in free ornithine suggests that this amino acid pool, or at least 97% of it, is not an intermediate in arginine synthesis. Other mechanisms for the formation of citrulline and arginine are put forward. Proline is apparently formed from glutamic acid. The results are generally at variance with the concept of amino acid families proposed by the Carnegie Institution group; with the possible exception of the glutamic acid family there is no evidence for the transfer of nitrogen from the family head to member amino acids. It is suggested therefore that these are really keto acid families and that transamination reactions are of major importance in amino acid biosynthesis from inorganic nitrogen.