Hydroxyl radical is a useful probe of the accessibility of the sugar moiety of nucleic acids to solvent. Here we compare the accessibility of free and ribosome-bound yeast tRNA(Phe), Escherichia coli tRNA(Phe), and E. coli tRNA(Leu2) to attack by hydroxyl radicals generated from Fe(2+)-EDTA. When bound to the P site of 30S ribosomal subunits, a discrete region, corresponding almost precisely to the anticodon stem-loop, is strongly protected; weaker protection is observed in the 3 strand of the D stem and in the variable loop. The protected nucleotides constitute a well-defined substructure, corresponding to the lower half of the anticodon-D loop coaxial arm of the tRNA crystal structure. This result suggests that the 30S P site contains a pocket that becomes inaccessible to the Fe(2+)-EDTA complex when tRNA is bound, whose minimum dimensions can be inferred from the boundaries of the protected region of tRNA. When bound to the P site of 70S ribosomes, the entire tRNA backbone becomes inaccessible to hydroxyl radicals. Since previous studies have shown that virtually the entire footprint of a P-site tRNA on 16S and 23S rRNAs is mimicked by the extremities of the tRNA (the anticodon stem-loop plus the 3 -terminal aminoacyl-pentanucleotide), protection of the entire tRNA was unexpected. We conclude that protection of the elbow of tRNA is due either to interactions with ribosomal proteins or to enclosure in an inaccessible site formed by association of the two ribosomal subunits.