There is evidence that the Escherichia coli polymerase recognizes and binds to three sites on the promoter DNA: the -10, -35, and -16 regions. Sequence homology was noted among the -10 sites (Pribnow box) and among the -35s with consensus sequences, TATAAT and TTGACA , respectively. Weak nucleotide sequence homology was detected at -16. Since the polymerase recognizes these sites in a multitude of promoters, one expects similarities in the three-dimensional structures. To date, no data directly bearing on such structures exist. Recently, Calladine ( Calladine , C.R. (1982) J. Mol. Biol. 161, 343-352) and, subsequently, Dickerson ( Dickerson , R.E. (1983) J. Mol. Biol. 166, 419-441) suggested "rules" for doublestranded DNA structures which were tested against data from several known crystals. Using these rules, I compare the deviations from "ideal" B-DNA of the twist angles, base pair roll, sideways shift, and propeller suppression in 56 promoters at the three sites. I also appended to these the twist angle computations on additional 77 promoters from the recently published compilation of promoter sequences. For the latter, additional nucleotides from the spacer regions were added. The results display similarities at the -10 site. Equally strong similarities were obtained for the -35 and the -16 contact regions. The existence of structural differences for some sites is likely to account for the different degrees of efficiency of the polymerase recognition and transcriptional regulation.