Catabolism of amino acids via the Ehrlich pathway involves transamination to the corresponding α-keto acids, followed by decarboxylation to an aldehyde and then reduction to an alcohol. Alternatively, the aldehyde may be oxidized to an acid. This pathway is functional in Saccharomyces cerevisiae, since during growth in glucose-limited chemostat cultures with phenylalanine as the sole nitrogen source, phenylethanol and phenylacetate were produced in quantities that accounted for all of the phenylalanine consumed. Our objective was to identify the structural gene(s) required for the decarboxylation of phenylpyruvate to phenylacetaldehyde, the first specific step in the Ehrlich pathway. S. cerevisiae possesses five candidate genes with sequence similarity to genes encoding thiamine diphosphate-dependent decarboxylases that could encode this activity: YDR380w/ ARO10, YDL080C/THI3, PDC1, PDC5, and PDC6. Phenylpyruvate decarboxylase activity was present in cultures grown with phenylalanine as the sole nitrogen source but was absent from ammonia-grown cultures. Furthermore, the transcript level of one candidate gene (ARO10) increased 30-fold when phenylalanine replaced ammonia as the sole nitrogen source. Analyses of phenylalanine catabolite production and phenylpyruvate decarboxylase enzyme assays indicated that ARO10 was sufficient to encode phenylpyruvate decarboxylase activity in the absence of the four other candidate genes. There was also an alternative activity with a higher capacity but lower affinity for phenylpyruvate. The candidate gene THI3 did not itself encode an active phenylpyruvate decarboxylase but was required along with one or more pyruvate decarboxylase genes (PDC1, PDC5, and PDC6) for the alternative activity. The Km and Vmax values of the two activities differed, showing that Aro10p is the physiologically relevant phenylpyruvate decarboxylase in wild-type cells. Modifications to this gene could therefore be important for metabolic engineering of the Ehrlich pathway.