We have investigated in detail the secondary and tertiary structures of the 16 S rRNA binding site of protein S8 using a variety of chemical and enzymatic probes. Bases were probed with dimethylsulfate (at A(N-1), C(N-3) and G(N-7)), with N-cyclohexyl-N'-(2-(N-methylmorpholino)-ethyl)-carbodiimide-p- toluenesulfonate (at G(N-1) and U(N-3)) and with diethylpyrocarbonate (at A(N-7)). The involvement of phosphates in hydrogen bonds or ion co-ordination was monitored with ethylnitrosourea. RNases T1, U2 and nuclease S1 were used to probe unpaired nucleotides and RNase V1 to monitor base-paired or stacked nucleotides. The RNA region, encompassing nucleotides 582 to 656 was probed within: (1) the complete 16 S rRNA molecule; (2) a 16 S rRNA fragment corresponding to nucleotides 578 to 756 obtained by transcription in vitro; (3) the S8-16 S rRNA complex; (4) the S8-RNA fragment complex; (5) the 30 S subunit. Cleavage or modification sites were detected by primer extension with reverse transcriptase. We present a three-dimensional model derived from mapping experiments and graphic modeling. Nucleotides in area 594-599/639-645 display unusual features: a non-canonical base-pair is formed between U598 and U641; and A595, A640 and A642 are bulging out of the major groove. The resulting helix is slightly unwound. Comparative analysis of probing experiments leads to several conclusions. (1) The synthesized fragment adopts the same conformation as the corresponding region in the complete RNA molecule, thus confirming the existence of independent folding domains in RNAs. (2) A long-range interaction involving cytosine 618 and its 5' phosphate occurs in 16 S rRNA but not in the fragment. (3) The fragment contains the complete information required for S8 binding. (4) The RNA binding site of S8 is centered in the major groove of the slightly unwound helix (594-599/639-645), with the three bulged adenines appearing as specific recognition sites. (5) This same region of the 16 S RNA is not exposed at the surface of the 30 S subunit.