Phe-tRNAPhe-C-C-A, Val-tRNAVal-C-C-A, and Ile-tRNAIle-C-C-A, which accept their amino acid on the 2'-OH of the 3'-terminal adenosine, are hydrolyzed readily by their aminoacyl-tRNA synthetase. If the 3'terminal adenosine in these tRNAs is replaced by either 3'-deoxyadenosine or formycin, little if any hydrolysis can be observed. Correspondingly Ser-tRNASer-C-C-A which accepts serine on the 3'-OH of the 3'-terminal adenosine is hydrolyzed by seryl-tRNA synthetase, whereas Ser-tRNASer-C-C-2'dA and Ser-tRNASer-C-C-F are not. Tyr-tRNATyr-C-C-A and all modified Tyr-tRNATyr-C-C-N, which can accept tyrosine on either the 2'OH or the 3'-OH of the 3'terminal adenosine, are not hydrolyzed by tyrosyl-tRNA synthetase. The data can be rationalized assuming that hydrolysis takes place only if the amino acid is bound to the nonaccepting OH and hence is not positioned at the amino acid binding site upon formation of the complex between aminoacyl-tRNA and aminoacyl-tRNA synthetase. In the formycin-carrying tRNA, the amino acid bound to the nonaccepting OH seems to be inaccessible to the enzymatic groups responsible for hydrolysis. Val-tRNAIle-C-C-3'dA and Ile-tRNAIle-C-C-3'DA cannot be hydrolyzed by isoleucyl-tRNA synthetase. Val-tRNAIle-C-C-A is hydrolyzed by the enzyme five times more rapidly than Ile-tRNAIle-C-C-A. Whereas Ile-tRNAIle-C-C-F is absolutely stable, Val-tRNAIle-C-C-F si hydrolyzed immediately. As shown by the earlier finding that valine misactivated by isoleucyl-tRNA synthetase cannot be permanently transferred to tRNAIle-C-C-A but to tRNAIle-C-C-3'dA, the 3'-OH is essential for preventing transfer of misactivated valine. It thus appears that valine is hydrolyzed off Val-tRNAIle-C-C-N if it is bound to the accepting 2'-OH in the binding site for isoleucine. A hypothesis is offered attempting to explain the experimental observations in mechanistic terms. We consider the hydrolytic action of the aminoacyl-tRNA synthetases as a general mechanism of "chemical proofreading" in the protein biosynthesis.