The specificity of binding of Watson - Crick base pairs by third strand nucleic acid residues via triple helix formation was investigated in a DNA pyrimidine triplex motif by thermal melting experiments. The host duplex was of the type A₁₀-X-A₁₀ : T₁₀-Y-T₁₀ , and the third strand T₁₀-Z-T₁₀ , giving rise to 16 possible triplexes with Z:XY inserts, 4 duplexes with the Watson-Crick base pairs (XY) and 12 duplexes with mismatch pairs (XZ), all of whose stabilities were compared. Two Z:XY combinations confirm the primary binding of AT and GC target pairs in homopurine-homopyrimidine sequences by T and C residues, respectively. All other Z:XY combinations in the T:AT environment result in triplex destabilization. While some related observations have been reported, the present experiments differ importantly in that they were performed in a T:AT nearest neighbor environment and at physiological ionic strength and pH, all of which were previously untested. The conclusions now drawn also differ substantially from those in previous studies. Thus, by evaluating the depression in Tm due to base triplet mismatches strictly in terms of third strand residue affinity and specificity for the target base pair, it is shown that none of the triplet combinations that destabilize qualify for inclusion in the third strand binding code for the pyrimidine triplex motif. Hence, none of the mismatch triplets afford a general way of circumventing the requirement for homopurine-homopyrimidine targets when third strands are predominated by pyrimidines, as others have suggested. At the same time, the applicability of third strand binding is emphasized by the finding that triplexes are equally or much more sensitive to base triplet mismatches than are Watson - Crick duplexes to base pair mismatches.