Updated publication reference for PubMed record(s): 30711543. Aminoacyl-tRNA synthetases (aaRSs), the enzymes responsible for coupling tRNAs to their cognate amino acids, minimize translational errors by intrinsic hydrolytic editing. Here, we compared the propensity of norvaline (Nva), a linear amino acid not coded for protein synthesis, to the proteinogenic, branched valine (Val) to mistranslate isoleucine (Ile) in proteins. We show that in the synthetic site of isoleucyl-tRNA synthetase (IleRS), aminoacylation and pre-transfer editing with Nva and Val occur at similar rates. Post-transfer editing was, however, more efficient with Nva as IleRS misaminoacylates Nva-tRNAIle at slower rate than Val-tRNAIle. Accordingly, an Escherichia coli strain lacking IleRS post-transfer editing misincorporated Nva and Val in the proteome to a similar extent and at the same Ile positions. However, Nva mistranslation inflicted higher toxicity than Val, in agreement with IleRS post-transfer editing domain being optimized for hydrolysis of Nva-tRNAIle. Furthermore, we found that the evolutionary related IleRS, leucyl- and valyl-tRNA synthetases (I/L/VRSs), all efficiently hydrolyze Nva-tRNAs even when editing of Nva seems redundant. Thus, we hypothesize that editing of Nva-tRNAs had already existed in the last common ancestor of I/L/VRSs, and that the editing domain of I/L/VRSs had primarily evolved to prevent infiltration of Nva into modern proteins.