In double-stranded DNA, tandem blocks of purines (Pu) and pyrimidines (Py) can form triplexes by pairing with oligonucleotides which also consist of blocks of purines and pyrimidines, using both Py.Pu.Py (Y-type) and Pu.Pu.Py (R-type) pairing motifs in a scheme called "alternate-strand recognition," or ASR [Jayasena, S. D., & Johnston, B. H. (1992) Biochemistry 31, 320-327; Beal P. A., & Dervan, P. B. (1992) J. Am. Chem. Soc. 114, 1470-1478]. We investigated the relative contributions of the Py.Pu.Py and Pu.Pu.Py blocks in the 16-bp duplex sequence 5'-AAGGAGAATTCCCTCT-3' paired with the third-strand oligonucleotides 5'-TTCCTCTTXXGGGZGZ-3' (XZ-16), where X and Z are either T or A and C is 5-methylcytosine, using chemical footprinting and get electrophoretic mobility shift measurements. We found that the left-hand, pyrimidine half (Y-block) of the third strand (TTCCTCTT, Y-8) forms a Py.Pu.Py triplex as detected by both dimethyl sulfate (DMS) probing and a gel-shift assay; in contrast, the triplex formed by the right-hand half alone (R-block) with X = T (TTGGGTGT, R-8) is not detectable under the conditions tested. However, when tethered to the Y-block (i.e., as XZ-16), the R-block contributes greatly increased specificity of target recognition and confers protection from DMS onto the duplex even under conditions unfavorable for Pu-Pu-Py triplexes (lack of divalent cations). In general, the 16-mer (XZ-16) can bind with apparent strength either greater or lesser than Y-8, depending on whether X and Z are A or T. The order of apparent binding strength, as measured by the target duplex concentration necessary to cause retardation of the third strand during gel electrophoresis, is TT-16 approximately AT-16 > Y-8 > AA-16 > TA-16. Chemical probing experiments showed that both halves of the triplex form even for AA-16, which binds with less apparent binding strength than the pyrimidine block alone (Y-8). The presence of the right half of the 16-mers, although detracting from affinity in cases of AA-16 and TA-16, provides strong specificity for the correct target compared to a target incapable of forming the Pu.Pu.Py part of the triplex. We discuss possible explanations for these observations in terms of alternate oligonucleotide conformations and suggest practical applications of affinity modulation by A-to-T replacements.