Accurate selection of amino acids is essential for faithful translation of the genetic code. Errors during amino acid selection are usually corrected by the editing activity of aminoacyl-tRNA synthetases such as phenylalanyl-tRNA synthetases (PheRS), which edit misactivated tyrosine. Comparison of cytosolic and mitochondrial PheRS from the yeast Saccharomyces cerevisiae suggested that the organellar protein might lack the editing activity. Yeast cytosolic PheRS was found to contain an editing site, which upon disruption abolished both cis and trans editing of Tyr-tRNA Phe . Wild-type mitochondrial PheRS lacked cis and trans editing and could synthesize Tyr-tRNA Phe , an activity enhanced in active site variants with improved tyrosine recognition. Possible trans editing was investigated in isolated mitochondrial extracts, but no such activity was detected. These data indicate that the mitochondrial protein synthesis machinery lacks the tyrosine proofreading activity characteristic of cytosolic translation. This difference between the mitochondria and the cytosol suggests that either organellar protein synthesis quality control is focused on another step or that translation in this compartment is inherently less accurate than in the cytosol.The aminoacyl-tRNA synthetases (aaRS) 2 are a ubiquitous and essential protein family required for protein synthesis (1-3). Structurally and functionally the aaRSs are divided into two unrelated but biochemically analogous groups, class I and class II (4 -6). The aaRSs attach amino acids to the 3Ј-ends of tRNA containing the corresponding anticodon sequence, and the resulting aminoacyl-tRNAs (aa-tRNAs) are used as substrates for ribosomal translation of mRNA. The accuracy of aa-tRNA synthesis is generally assured by the existence of aaRSs specific for each particular amino acid-tRNA pair. Cognate tRNA recognition and discrimination of non-cognate RNAs are achieved by sequencespecific direct and indirect readout of the numerous combinations of bases present in tRNAs (7-10). The relative structural simplicity and inherent similarity between the amino acid substrates makes their accurate recognition and discrimination more challenging. Although some amino acids such as cysteine and tyrosine are different enough to allow their specific recognition by a particular aaRS (11, 12), others such as valine and isoleucine are less easily distinguished. For example the class I aaRS isoleucyl-tRNA synthetase (IleRS) is only able to poorly discriminate against valine, which has a misactivation rate of about 1:200 compared with the cognate substrate isoleucine. Despite this significant rate of misactivation and misaminoacylation, the accuracy of translation is not compromised because of the existence of an intrinsic proofreading and editing mechanism in IleRS that specifically hydrolyzes both misactivated Val-AMP and misaminoacylated Val-tRNA Ile (13,14). In addition to IleRS, many other class I and class II aaRSs also employ editing to prevent release of non-cognate aa-tRNA and subsequent l...