We report on the determination of the solution structure of two sequence-related oligonucleotides, d(GTACGTAC)2 and d(CATCGATG)2. Results have been obtained by using a combined approach of (a) two-dimensional NMR, including proton and phosphorus experiments, (b) restrained molecular mechanics performed with sugar phase angle, backbone epsilon angle, and NOE distances as input, and (c) back-calculation refinements against the NOE spectra at various mixing times. The two oligonucleotides adopt the B-DNA structure with, however, noticeable differences centered on their core sequence and especially the CpG step. Due to the permutation of its flanking residues, the CpG step modifies its twist values and backbone epsilon value; globally, the CpG step appears more flexible within the tetranucleotide TCGA than ACGT. The solution structure of d(GTACGTAC)2 differs from the previously reported X-ray structure, which was found to be A-form throughout [Takusagawa, F. (1990) J. Biomol. NMR 3, 547-568]. On the other hand, in the X-ray structure of d(CCAACGTTGG)2 [Privé et al. (1991) J. Mol. Biol. 217, 177-199] the structure of the ACGT sequence is similar to that found in solution d(GTACGTAC)2. Similarly, the central TCGA tetranucleotide of d(CATCGATG)2 presents a solution structure analogous to that observed on the X-ray structures of both d(CGATCGATCG)2 [Grzeskowiak, et al. (1991) J. Biol. Chem. 266, 8861-8883] and d(CGATCGmeATCG)2 [Baïkalov, et al. (1993) J. Mol. Biol. 231, 768-784]. At the end we discuss the possible biological significance of the particular structures exhibited by the ACGT and TCGA tetranucleotides.