The phenylalanyl-tRNA synthetases from Escherichia coli, Saccharomyces cerevisiae, Neurospora crassa, and turkey liver activate a number of phenylalanine analogues (tyrosine, leucine, methionine, p-fluorophenylalanine, beta-phenylserine, beta-thien-2-ylalanine, 2-amino-4-methylhex-4-enoic acid, mimosine, N-benzyl-L- or N-benzyl-D-phenylalanine, and ochratoxin A), as demonstrated by Km and kcat of the ATP/PPi pyrophosphate exchange. Upon complexation with tRNA, the enzyme-tRNAPhe complexes show a significantly increased initial discrimination of these amino acid analogues expressed in higher Km and lower kcat values, as determined by amino-acylation of tRNAPhe-C-C-A(3'NH2). The overall accuracy is further enhanced by a second discrimination, a proofreading step. The strategies employed by the enzymes with respect to accuracy differ. Better initial discrimination in the aminoacylation and less elaborated proofreading for the E. coli enzyme can be compared to a more efficient proofreading by other synthetases. In this way the comparatively poor initial amino acid recognition in the case of the S. cerevisiae and N. crassa enzymes is balanced. The extent of initial discrimination is therefore inversely coupled to the hydrolytic capacity of the proofreading. A striking difference can be noted for the proofreading mechanisms. Whereas the enzymes from E. coli, S. cerevisiae, and N. crassa follow the pathway of posttransfer proofreading, namely, enzymatic hydrolysis of the misaminoacylated tRNA, the turkey liver enzyme uses tRNA-dependent pretransfer proofreading in the case of natural amino acids. In spite of the same subunit structure and similar molecular weight, the phenylalanyl-tRNA synthetases from a prokaryotic and lower and higher eukaryotic organisms show obvious mechanistic differences in their strategy to achieve the necessary fidelity.