We have investigated in detail the higher-order structure of 16 S ribosomal RNA, both in its naked form and in 30 S ribosomal subunits. Each base in the 16 S rRNA chain has been probed using kethoxal (which reacts with guanine at N1 and N2), dimethylsulfate (which reacts with adenine at N1 and cytosine at N3) and 1-cyclohexyl-3-(2-morpholinoethyl)-carbodiimide metho-p-toluenesulfonate (which reacts with uracil at N3 and guanine at N1). The sites of reaction were identified by primer extension with reverse transcriptase using synthetic oligodeoxynucleotide primers. These results provide a detailed and rigorous experimental test of a model for 16 S rRNA secondary structure, which was derived mainly from comparative sequence analysis. Our data also provide information relevant to tertiary and quaternary structure of 16 S rRNA. Data obtained with naked 16 S rRNA show reasonably close agreement with the proposed model, and data obtained with 30 S subunits show nearly complete agreement. Apart from an apparent overall "tightening" of the structure (in which many weakly reactive bases become unreactive), assembly of the proteins with 16 S rRNA to form 30 S subunits brings about numerous local structural rearrangements, resulting in specific enhancements as well as protections. In many instances, the ribosomal proteins appear to "tune" the 16 S rRNA structure to bring it into accordance with the phylogenetically predicted model, even though the RNA on its own often seems to prefer a different structure in certain regions of the molecule. Extensive protection of conserved, unpaired adenines upon formation of 30 S subunits suggests that they play a special role in the assembly process, possibly providing signals for protein recognition.